Industries / Energy, water and waste / Power generation / Small modular reactors
Waste
Waste work plans what a small reactor's spent fuel and other radioactive leftovers will be, and how they are stored, moved and finally disposed of.
5
Milestones achieved, from 1994 to 2024.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1994
Stored salt fuel makes gas
The Molten Salt Reactor Experiment at ORNL ran on fuel dissolved in molten fluoride salt from June 1965 to December 1969. At shutdown about 4,710 kg of salt, holding about 36.3 kg of uranium, was drained into two tanks and left to freeze. Radiation kept breaking down the salt. In April 1994 a gas sample showed uranium had moved out of the tanks into the off-gas pipes, about 2.6 kg of it had collected in a charcoal bed, and gas samples held fluorine at 350 mm of mercury and uranium hexafluoride at 70 mm. ORNL had to treat the build-up as a criticality risk, the chance of an unplanned chain reaction, and design a system to pump the gas out.
Lessonsalt fuel keeps changing after shutdown, so it needs a stable waste form, not only a sealed tank.
Credit
ORNL (D. F. Hollenbach and C. M. Hopper; J. C. Rudolph, G. D. Del Cul and J. Caja)
2014
Pebbles outlive their store
Germany's AVR test reactor at Jülich left about 300,000 spent fuel pebbles, held in 152 Castor casks. The first cask was loaded on 11 August 1993, and the store was licensed until 2013. Renewal needed proof of earthquake safety that could not be supplied in time, so in 2014 the state of North Rhine-Westphalia ordered the store emptied without delay, and the pebbles have stayed there without a licence since 1 July 2013. On 2 March 2026 the Higher Administrative Court of Berlin-Brandenburg cleared their transport to the store at Ahaus, citing the public interest in emptying a store "whose licence expired years ago".
Lessoninterim storage needs its next destination settled before its licence runs out.
Credit
Jülicher Entsorgungsgesellschaft für Nuklearanlagen (JEN, the operator); North Rhine-Westphalia's nuclear regulator; the Higher Administrative Court of Berlin-Brandenburg
2022
Small cores, more waste
In the journal PNAS, Lindsay Krall, Allison Macfarlane and Rodney Ewing compared three SMR designs, NuScale's water-cooled module, Toshiba's sodium-cooled 4S and Terrestrial Energy's molten-salt IMSR, with a 1,100 MWe pressurized water reactor. They found the SMRs would raise the volume of waste needing management and disposal by factors of 2 to 30, mainly because small cores leak more neutrons, and that NuScale's spent fuel per unit of energy would be 1.7 times a large reactor's. NuScale's chief technology officer, Jose Reyes, wrote to the journal that the study used NuScale's older 160 MW thermal core rather than its 250 MW core, whose burnup he said matches today's fleet. The authors replied that the choice was deliberate, because only the 160 MW design had been filed with the NRC.
Lessona waste comparison holds only for the design version and burnup it assumes.
Credit
Krall, Macfarlane and Ewing, Stanford University's Center for International Security and Cooperation
2024
New spent fuel, old casks
Staff at Idaho and Pacific Northwest national laboratories checked, using only public data, whether spent fuel from new designs fits today's storage and transport systems. Pebble fuel from X-energy's Xe-100 and Kairos Power's KP-X fit, with limits on canister width to rule out a chain reaction, and reached only about 151 °C, against about 375 °C for today's spent fuel. For spent molten-salt fuel, they found the stable waste form and the steps to make it are "major knowledge gaps".
Credit
G. M. Petersen, U. Carvajal Nunez and A. M. Clarke Newman (INL) and B. Hanson (PNNL), for DOE's Spent Fuel and Waste Science and Technology program
2024
Volume is the wrong measure
Haruko Wainwright and colleagues at MIT linked three open-source codes, OpenMC for the reactor, NWPY for the size of the repository and PFLOTRAN for groundwater flow, and compared five kinds of small reactor with a large AP1000. TRISO-fuelled reactors made 25 to 30 times the reference's spent fuel volume, yet needed less repository space and gave lower peak doses; a small water-cooled reactor made 3.3 times its spent fuel mass. Their review traced earlier contradictory studies, including Krall's, largely to different assumed burnups and different ways of counting volume. They published their code and inputs.
Lessonwaste studies agree only when they share inputs and definitions.
Credit
Haruko M. Wainwright and co-authors, MIT
Watch for
Recycling for small reactors
Oklo plans a plant in Oak Ridge, Tennessee, to recycle used fuel from today's reactors into metal fuel for its Aurora fast reactors, by pyroprocessing, a dry method that separates the fuel with electric current in molten salt. It is the first phase of a $1.68 billion fuel center. Oklo says it has agreed a licensing plan with the NRC and is in pre-application talks, expects to make fuel by the early 2030s, and claims recycling shortens the time used fuel must be stored "from tens of thousands of years to a few hundred".
Achieved whenthe NRC licenses the plant, it makes fuel from used commercial fuel, and someone outside Oklo publishes the volume and lifetime of the waste that remains.
Claimed by
Oklo, September 2025, repeated in a 2026 fact sheet
Open
01
Waste figures for licensed designs
Independent, open estimates of spent fuel mass, volume and decay heat per unit of energy for small reactor designs now in licensing, starting with NuScale's 250 MW thermal core and GE Vernova Hitachi's BWRX-300, run in MIT's open R2R4SNF framework with every input traced to a public document.
The briefCredit · unclaimed02
Predict gas from stored salt
An open model that predicts both how much fluorine and uranium hexafluoride gas builds up above frozen fuel salt in storage and when it starts to escape, tested against the measurements ORNL made at the Molten Salt Reactor Experiment.
The briefCredit · unclaimed
Open 01 · the brief
Waste figures for licensed designs
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
Independent, open estimates of spent fuel mass, volume and decay heat per unit of energy for small reactor designs now in licensing, starting with NuScale's 250 MW thermal core and GE Vernova Hitachi's BWRX-300, run in MIT's open R2R4SNF framework with every input traced to a public document.
Why it is still open
Krall and colleagues analysed NuScale's older 160 MW thermal core, said the 250 MW core could be analysed once its design data were available, and still disagree with NuScale about what it would show. MIT's framework compared five generic reactor types, such as "small PWR", not named, licensed designs.
Built with what exists
R2R4SNF from GitHub (github.com/hmwainw/R2R4SNF), which runs OpenMC, NWPY and PFLOTRAN; NuScale's public answers to the National Academies (2021) and public NRC and Canadian licensing documents; one workstation.
First step
Rerun the framework's published small-PWR case, then swap in the 250 MW thermal core's public parameters.
Done looks like
Results for at least two licensed designs, every input cited to a page, set beside the framework's AP1000 and generic small-PWR cases, with a second group rerunning the inputs to within 5%.
Open 02 · the brief
Predict gas from stored salt
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open model that predicts both how much fluorine and uranium hexafluoride gas builds up above frozen fuel salt in storage and when it starts to escape, tested against the measurements ORNL made at the Molten Salt Reactor Experiment.
Why it is still open
In 1996 ORNL projected from assumed yields that 300 to 650 mol of fluorine could have formed by 1994, two to five times the amount identified in the off-gas system, and wrote that without a better understanding of how the gas forms "it is impossible to construct a precise time line" for it. The 2024 INL and PNNL review says the extent of gas release from solidified salt "requires assessment beyond what is presented herein". New production rates exist: a TU Delft team irradiated fluoride salts at Petten and reported G-values, the number of fluorine molecules made per 100 eV of radiation absorbed, of about 0.004 to 0.021. A search of DOE's OSTI library and the web found no later model checked against the MSRE record.
Built with what exists
ORNL's MSRE reports on OSTI (salt inventory, the 1969 drain, the 1994 gas samples); the TU Delft G-values; OpenMC's depletion and decay tools to estimate the radiation dose inside the salt over time; Python.
First step
Gather every gas measurement from the MSRE storage period in ORNL's OSTI reports into one dated table.
Done looks like
A published model that, from the salt's 1969 inventory and storage temperatures, predicts the amount of fluorine found by 1994 within a factor of two and the year gas release began within five years, with code and inputs released and rerun by a second group.