Rian Bahran, Ph.D.

New open-source tool for the nuclear and AI communities! NRC-ADAMS-AI, by Chris Walk. The U.S. Nuclear Regulatory Commission’s Agencywide Documents Access and Management System (ADAMS) contains decades of licensing actions, safety evaluations, and inspection reports. Yet integrating this public corpus into modern #GenAI and #agenticAI workflows has remained a challenge. Download terabytes and finetune/RAG? To help address this gap, the Advanced Energy Systems Laboratory (AESL) at Texas A&M University has released an open-source MCP tool that enables agent-based systems to access and reason over ADAMS documents directly within your AI pipelines, transactionally! Researchers, developers, and analysts can now incorporate regulatory and technical context from ADAMS into their models with far less friction. 🔗 Tool and documentation: https://lnkd.in/gRFDGVnh 🔗 Learn more about what we’re doing in AESL: https://lnkd.in/gZNBHqqR This work is supported by the U.S. Department of Energy / National Nuclear Security Administration under Award Number DE-NA0003921, with a special thanks to the ETI2.0 consortium (https://eti2.gatech.edu/) for its collaborative environment and support. Chris Walk is our research intern and Computer Science, Class of 2027, Georgia Tech, who put this tool together. Dan Watson, PhD Candidate and researcher leading project. And we are just getting started—there are many exciting developments to come at the intersection of nuclear engineering and AI. #GenAI #ArtificialIntelligence #AgenticAI #Nuclear #NuclearEngineering #NuclearEnergy #NuclearRegulation #NRC #OpenSource #AI #Energy

43

2 Comments

My scepticism of AI's current abilities continues but others in the nuclear industry are pushing ahead. This report is a worth a read to understand the variety of applications, stakeholders and their motivations, and possible risks of its use in nuclear. It's adding a lot to my "need to know more about this" list.

9

1 Comment

Small Modular Reactors (SMRs) are reshaping the global nuclear landscape — and Artificial Intelligence is about to reshape the rules behind them. SMR programmes are expanding rapidly in the U.S., Canada, China and Russia, while Europe debates its path and development banks explore how SMRs could bring electricity to regions like sub-Saharan Africa, where half a billion people still live without power. But as deployment accelerates, one challenge grows faster than the reactors themselves: licensing and regulatory workload. Nuclear energy is subject to one of the most demanding international oversight regimes in the world. The volume of documentation, reviews and safety assessments is reaching a point where it risks overwhelming even the International Atomic Energy Agency (IAEA). And this is where AI steps in. This month, during the IAEA’s first International Symposium on AI and Nuclear Energy, the agency signed a landmark agreement with Atomic Canyon — a U.S. company using AI to process nuclear documentation, accelerate licensing, and build high-fidelity simulations that enhance safety while reducing timelines. Westinghouse Electric Company, Google and other players are moving in the same direction, developing AI tools that cut component costs, support design decisions and give engineers “Iron Man-level” access to regulatory and technical data. The implications go far beyond efficiency. AI could redefine global nuclear competitiveness and shift geopolitical balance. Historically, the U.S. has shaped international nuclear standards. But China’s rapid reactor build-out is changing that equilibrium. If AI becomes the engine that drives licensing, operations and safety cases, the countries that lead in AI-nuclear integration could set the next generation of global norms. The future of nuclear energy will not be determined only by who builds more reactors, but by who can deploy, license and operate them faster, safer and smarter — and AI is becoming the decisive factor. #NuclearEnergy #SMR #ArtificialIntelligence #IAEA #EnergyTransition #Innovation #Decarbonization #Geopolitics

51

1 Comment

One of the top priorities for fusion energy science is to develop an understanding of power exhaust sufficient to inform the engineering design of the systems that handle that intense power. In the new work below, Bob Wilcox and colleagues utilized the DIII-D National Fusion Facility to generate a comprehensive experimental data set across different power exhaust geometries and then coupled multiple models to achieve accurate calculations of plasma behavior. In a tokamak fusion plasma, the far edge establishes a pedestal, which is a steep cliff connecting the highly confined hot and dense plasma to the unconfined region that interacts with the device wall. The height of this pedestal largely determines the power output of the plasma, so pedestal modeling of future devices must be accurate in order to hit the targeted output. As the results show, simply coupling different codes together does not get the job done. The research team identified optimizations that correctly describe the reality observed in experiments. This is particularly important because the simple integration of codes produces pedestal heights higher than reality, which would lead to overestimates of fusion power output. This work brings together researchers from Fusion and Fission at ORNL, Type One Energy, Commonwealth Fusion Systems, General Atomics, and Lawrence Livermore National Laboratory. R.S. Wilcox, et al., Nucl. Fusion 66, 036005 (2026), https://lnkd.in/gySGZGsQ #fusionenergy #science #computationalscience

71

Great news: Radiation and Nuclear Safety Authority of Finland (STUK) concluded that Posiva can start the production of the tunnel backfill for the first disposal tunnel. 👏

38

𝗟𝗔𝗠𝗣𝗥𝗘 (𝟭𝟵𝟱𝟳) — 𝗣𝗵𝘆𝘀𝗶𝗰𝘀 𝗦𝗮𝗶𝗱 𝗬𝗲𝘀. 𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻𝘀 𝗦𝗮𝗶𝗱 𝗔𝗯𝘀𝗼𝗹𝘂𝘁𝗲𝗹𝘆 𝗡𝗼𝘁.(𝗙𝗼𝗿𝗴𝗼𝘁𝘁𝗲𝗻 𝗥𝗲𝗮𝗰𝘁𝗼𝗿𝘀 𝗘𝗽𝗶𝘀𝗼𝗱𝗲 #𝟴𝟭) Most reactors are designed to prevent fuel from melting. The Los Alamos Molten Plutonium Experiment (LAMPRE) began with the assumption that it already had. In the early 1950s, Los Alamos engineers asked a question: If meltdown is the failure mode… what if the fuel is already liquid? LAMPRE construction began in the mid-1950s and went critical in 1957. This was a fast-spectrum experimental reactor designed to test molten plutonium alloy as fuel. Not uranium oxide pellets. Not ceramic pins. MOLTEN PLUTONIUM. Pure plutonium melts at about 640°C . LAMPRE used a plutonium-iron alloy, lowering the melting point into a ~500°C range. Still hotter than a commercial oven. Still chemically aggressive. The fuel was sealed inside tantalum capsules with liquid sodium as the coolant. On paper, it had a certain brutal logic; if the fuel is already molten, it cannot melt down. The plutonium came from the U.S. production reactor complex — weapons-grade material alloyed and heated until it flowed like dense metallic syrup. The idea looked great on paper. And then operations began. Plutonium is not a polite metal. Even alloyed, it attacked structural materials. VIGOROUSLY. Tantalum was used because almost everything else degraded faster. Corrosion wasn’t a maintenance task. It was a race against time. Leakage risk carried a different character altogether. You are containing dense, alpha-emitting, fissile liquid metal at ~500°C. A failed capsule isn’t a cracked pellet. It is liquid plutonium where it does not belong — like trying to keep a drop of mercury still on a vibrating steel table, except this mercury can sustain a chain reaction. Freezing posed its own hazard. If temperature dropped, the fuel solidified inside its capsule. Restarting required carefully reheating a core full of frozen fissile metal. Think restarting an engine after the oil turned to metal. Remote handling was mandatory. Gloveboxes. Shielded hot cells. Exacting contamination control. Every gram mattered. LAMPRE operated intermittently from 1957 to 1963. A larger follow-on concept, LAMPRE II, was studied but never scaled. Because by the early 1960s, something became clear; yes, molten plutonium fuel was technically feasible. But feasibility is not the same thing as operational survivability. - The materials challenges. - The contamination risk. - The handling complexity. - The razor-thin tolerance for error. Physics may permit it. Operations must survive it. LAMPRE did not explode. It did not melt down. It did not become a headline disaster. It simply answered the question; can we build a reactor that uses molten fuel? Physics said yes. Operations said absolutely not. #NuclearHistory #LessonsLearned

129

7 Comments

Nuclear power is now firmly positioned at the center of U.S. energy strategy under POTUS. That commitment was underscored on Friday, when President Trump signed four executive orders aimed at accelerating the deployment of next-generation nuclear technologies. These directives address the full nuclear development cycle—from funding and licensing to testing and commercialization—with a bold goal of quadrupling America’s nuclear capacity by 2050. A key example of this momentum is the exclusive global partnership between NuScale Power and ENTRA1 Energy. Under this agreement, ENTRA1 Energy has the rights to develop, manage, own, and operate energy production plants powered by NuScale’s NRC-approved Small Modular Reactor (SMR) technology. Through this joint venture, ENTRA1 Energy Plants™ will deliver immediate, scalable, and clean baseload power to meet the rising energy demands of hyperscalers and AI infrastructure—supporting U.S. energy independence and digital resilience. This strategic partnership is the only US NRC approved technology and development platform ready for deployment 🚀 #NuScale #ENTRA1 #SMR #Nuclear

69

Hot off the press! 🧑‍🍳🥧 My team at the Nuclear Scaling Initiative just released a report identifying key nuclear energy supply chain bottlenecks—and how to address them to enable large-scale deployment.

34

Three levers that matter in nuclear: waste, physics, infrastructure. When a fuel concept addresses all three at once, it deserves attention.

21

Modern licensing of safety-critical systems hinges on high-fidelity models (CFD, neutronics, multiphysics). Yet, regulatory confidence now demands more than a best-estimate answer—it requires a rigorous, defensible treatment of both epistemic and aleatory uncertainties. The solution is a foundational framework combining Uncertainty Quantification (UQ) with Verification, Validation, and Accreditation (VV&A). This mathematical backbone supports robust, risk-informed regulatory decisions. What are the key sources of uncertainty we must quantify? • Parametric: Material properties, boundary conditions, operational ranges. • Model-form: Simplifications, closure laws, physical assumptions. • Numerical: Discretization error, solver convergence. • Experimental: Measurement noise, scaling effects. Core UQ Techniques Enabling This: ✔ Advanced sampling (Monte Carlo, Latin Hypercube) ✔ Bayesian inference for calibration ✔ Polynomial chaos expansions & surrogate modeling ✔ Global sensitivity analysis (Sobol indices, adjoint methods) This process propagates uncertainty to critical Licensing Figures of Merit: → Peak cladding temperature → Reactivity margins → Structural failure probabilities → Core damage frequency (CDF) In nuclear licensing, this has driven the industry shift from conservative bounding analyses toward Best Estimate Plus Uncertainty (BEPU) methodologies, now explicitly endorsed by major regulators. Regulatory focus is now on: Quantified confidence intervals – not just point estimates. Complete traceability of modeling choices and assumptions. Integrated frameworks coupling deterministic simulations with PRA. Explicit margin characterization under uncertainty. The result? Enhanced safety realism and economic efficiency—especially for advanced reactors where outdated conservatisms are no longer justified. The Future is Integrated: As computational power grows and digital twins become operational realities, UQ is evolving from a final check into a core, integrated component of the entire design, safety analysis, and licensing workflow. The message is clear: The future of technical licensing is not deterministic. It is probabilistic, uncertainty-aware, and mathematically rigorous. I'm interested in your perspective: · What is the most significant practical challenge your team faces in implementing a full BEPU/UQ framework? · Where have you seen the most impactful safety or economic benefit from a risk-informed approach? #UncertaintyQuantification #UQ #RiskInformed #NuclearSafety #BEPU #Licensing #VerificationAndValidation #VV&A #Engineering #AdvancedReactors #DigitalTwin #PRA #RegulatoryScience

12

Nuclear in the US is growing strong. With progress towards faster deployments, the barrier to entry for nuclear power is seeming more and more attainable. This is evidenced by many nuclear companies coming to the US to implement the use of their technologies. Read the article by the Nuclear Advocacy Resource Organization (NARO) to learn more! #innovation #energy #nuclear #technology #smr #future #sustainability #policy #datacenters

9

The OECD Nuclear Energy Agency (NEA) has published a report on the potential for replacing existing coal-fired power plants with Small Modular Reactors (SMRs). A webinar will be held soon to present and discuss the findings of this report. I am pleased to share that I had the opportunity to contribute a section to this report, focusing on Korea’s decarbonization strategy and nuclear transition.

16