The Figure of Merit Problem in Proposed Part 57: Why the SNM Mass Limit Is the Wrong Tool for Reactivity Control

May 21, 2026

Executive Summary

Proposed 10 CFR Part 57 establishes two eligibility criteria for entry into its streamlined licensing framework. The first — a 1 rem TEDE dose limit at the unrestricted area boundary — is a well-constructed Figure of Merit that the NRC itself describes as such in the proposed rule's preamble, and it already does the work the second criterion is trying to do, and does it better. The second — a 10 metric ton limit on special nuclear material (SNM) inventory — is a defensible but imprecise proxy for one classical safety function, and a poor proxy for another. The good news is that Part 57 already contains the right criterion for reactivity control: the design attribute demonstration required under § 57.30. It just hasn't been elevated to entry criterion status. With the Part 57 comment period closing June 15, 2026, this is the technical argument worth making — and the practitioners fluent in both DOE and NRC regulatory space are the ones best positioned to make it.

Two Regulatory Frameworks, One Shared Safety Logic

The argument in this article draws on concepts that appear in both the DOE and NRC regulatory frameworks, and it is worth being explicit about the provenance of each.

The four classical safety functions that every nuclear reactor design must fulfill originate in DOE regulatory space. DOE-STD-1271-2025, Authorization Pathways for Nuclear Facilities, codifies them as: Radioactive Material Confinement, Nuclear Reactivity Control, Fission and Decay Heat Removal, and Preservation of adequate radiation shielding. In plain terms:

1. Reactivity control — The design must be able to shut the reactor down and keep it shut down under all credible conditions. If you cannot control reactivity, you cannot prevent a power excursion.

2. Fission and decay heat removal — Even after shutdown, a reactor continues to generate heat from radioactive decay. If that heat cannot be removed, fuel can be damaged and radionuclides can be released.

3. Confinement of radionuclides — If the first two functions fail, physical and functional barriers must prevent radioactive material from reaching the public.

4. Shielding — Radiation emitted during normal operation and transients must be attenuated to protect workers and the public.

The NRC does not use the DOE's term "Fundamental Safety Functions" as a defined regulatory term. But the Part 57 proposed rule reflects the same underlying logic: its six design criteria attributes — reactivity control, heat removal, fission product retention, shielding, radioactive effluents control, and security by design — map directly onto these four classical functions. Practitioners who work across both DOE authorization and NRC licensing space recognize the shared architecture immediately, even when the vocabulary differs.

The concept of a Figure of Merit, by contrast, is used in NRC space — specifically, in the Part 57 proposed rule's own preamble. The NRC states directly: "While the dose-based entry criterion would be computed in terms of dose, it is a figure of merit used to characterize the minimum requirements for design, fabrication, construction, testing, operational limits, and performance for safety-related SSCs." This language has a long lineage in NRC siting criteria under § 50.67. I extend the same analytical framework in this article to the SNM mass limit — not because the NRC calls it a figure of merit, but because evaluating both entry criteria through that lens is the clearest way to expose the technical mismatch the NRC is asking commenters to address.

The 1 rem TEDE Dose Criterion: A Well-Constructed Figure of Merit That Already Covers More Ground Than It Gets Credit For

The dose-based entry criterion in proposed § 57.25(a) requires applicants to demonstrate that a bounding accident does not result in a dose exceeding 1 rem TEDE to a member of the public in the unrestricted area. The NRC explicitly calls this a figure of merit in the preamble, and the characterization is well earned. It maps directly and completely to the confinement of radionuclides safety function and, through it, to shielding and effluent control. The analysis required to demonstrate compliance — a source term calculation combined with radiological consequence analysis — directly answers the question that safety function is asking: if the worst credible accident occurs, does the design prevent unacceptable radiological harm to the public?

Crucially, a credible worst-case accident analysis is not independent of the heat removal safety function. To bound the source term, the analyst must make assumptions about fuel damage, which depends on whether decay heat is successfully removed. A design that cannot passively remove decay heat will produce a larger source term and will have more difficulty meeting the 1 rem TEDE criterion. The dose criterion therefore already exerts indirect pressure on the heat removal safety function without requiring a separate inventory-based screen.

The dose criterion is also technology-inclusive in precisely the way a figure of merit for a technology-inclusive rule should be. A molten salt reactor, a gas-cooled reactor, a metal-fueled fast reactor, and a small PWR all answer the same question using the same metric, even though the source terms, transport pathways, and barrier configurations are entirely different. The figure of merit does not prejudge the answer — it sets the performance target and allows the design to meet it through whatever physical means are appropriate to the technology.

The 10 Metric Ton SNM Limit: Applying the Same Framework Reveals the Mismatch

The fuel mass limit in proposed § 57.25(b) caps the total inventory of special nuclear material — the uranium, thorium, and plutonium in the reactor — at 10 metric tons per unit. The NRC does not call this a figure of merit in the preamble. Applying that analytical lens to it, as I do here, is my extension of the framework — and the extension reveals why the limit is doing two different jobs with unequal competence.

For the heat removal safety function, the mass limit has real, if imprecise, logic behind it. More fissile and fertile material generally means more fission products, more decay heat at shutdown, and a greater demand on passive heat removal systems. A cap on total inventory therefore provides an indirect bound on the peak decay heat load the plant must manage without active cooling. It is not a precise criterion — the actual decay heat depends on power history, fuel composition, and burnup as well as total mass — but it is a defensible proxy. The more important point, developed above, is that the 1 rem TEDE dose criterion already captures this pressure on the heat removal safety function indirectly through the source term analysis. The mass limit is adding an imprecise redundancy to a screen that is already doing the work.

For the reactivity control safety function, the mass limit fails more fundamentally — and here a fair objection must be addressed before the argument can proceed.

Preempting the Obvious Objection: Correlation Within a Fixed Design Is Not the Right Standard for a Technology-Inclusive Rule

A reasonable reviewer might object that for a given fuel form, enrichment, and geometry, total SNM mass is monotonically correlated with reactivity — more mass means more neutrons and a higher multiplication factor. This is true within a fixed design. But Part 57 is explicitly technology-inclusive, and the mass limit is applied as an entry criterion across all eligible designs regardless of technology. In that cross-technology context, the correlation breaks down.

Reactivity — the condition that determines whether a chain reaction will grow, remain steady, or die out — is governed by the neutron multiplication factor. That factor is a function of geometry, enrichment, moderation, neutron reflection, and coolant density, not mass alone. Two reactor designs with identical SNM inventories can have radically different reactivity characteristics depending on how the fuel is arranged, what surrounds it, and what moderates or reflects the neutrons. A design using TRISO fuel in a graphite moderator behaves fundamentally differently from one using metal fuel in a sodium coolant or dissolved fissile material in molten salt — even at identical inventory levels.

More importantly: a design can be inherently and passively subcritical at a mass level that would be supercritical in a different geometric or moderating configuration. The mass limit screens on a variable that carries information about nuclear hazard within a fixed technology type but not across technologies. For an entry criterion in a technology-inclusive rule, that is a fundamental mismatch between the metric and the safety function it nominally represents.

The Right Criterion for Reactivity Control Is Already in the Rule — It Just Needs to Be Elevated

The stronger observation is not that the NRC needs to invent a new criterion for reactivity control. It is that the rule already contains one.

Proposed § 57.30 requires applicants to demonstrate a set of design criteria attributes, including reactivity control. Meeting that attribute necessarily involves demonstrating that the net reactivity coefficient — the rollup of temperature, moderator, void, and any other relevant feedback mechanisms specific to the technology — is negative across the operating cycle within the bounds of the safety analysis.

This is exactly the calculation that a reactivity control criterion should require. In my earlier work performing reload design and analysis for large LWR fuel cycles — calculating temperature, moderator, and void coefficients cycle by cycle, then confirming that the net coefficient was negative across the entire operating cycle — this is precisely what the reactivity control analysis produced. It is worth being clear about the standard, because it is easy to state it incorrectly: the criterion is not that every individual coefficient must be negative at every point in the cycle. The B&W 177 FA design, for instance, can exhibit a briefly positive moderator coefficient at approximately 4 effective full-power days into the cycle due to the relatively high concentrations of boric acid at the beginning of cycle, yet the conglomerate net reactivity coefficient remained negative throughout, providing the required safety assurance. The technically correct standard — and the one that should appear in a reactivity control entry criterion — is that the net coefficient is negative across the operating cycle within the bounds of the safety analysis.

That analysis is not analytically expensive. It is a standard nuclear physics calculation that any competent nuclear analyst can perform early in the design process using established neutronics tools. It is technology-inclusive: every reactor technology has a calculable set of reactivity feedback mechanisms, and the net coefficient requirement applies regardless of whether those mechanisms involve Doppler broadening in a uranium oxide lattice, graphite temperature feedback in a gas-cooled design, sodium void reactivity in a fast reactor, or salt density changes in a molten salt concept.

Critically, elevating this demonstration to entry criterion status adds no new analytical burden beyond what § 57.30 already demands. The work is already required. The proposal is simply to make the timing of the demonstration explicit — at eligibility screening under § 57.25, not only at the design attribute stage under § 57.30 — and to tie it directly to the reactivity control safety function the mass limit was never equipped to represent.

What a Strong Comment on Q1-1 Should Argue

The NRC's specific question is: in lieu of the deterministic material limit, should the Commission consider an alternative performance-based entry criterion?

A technically grounded comment should make three connected points.

First, the 1 rem TEDE dose criterion — which the NRC itself calls a figure of merit — already indirectly captures the heat removal safety function through the source term analysis. The mass limit adds an imprecise redundancy to a screen that is already doing the relevant work, and the NRC's own draft Part 57 preamble language supports this reading.

Second, applying the same figure of merit framework to the mass limit reveals that it fails as a reactivity control criterion across a technology-inclusive rule. Mass is not the controlling variable for reactivity when technology type is not fixed. The relevant physics — geometry, enrichment, moderation, and feedback coefficients — are not captured by an inventory count.

Third, and most constructively: the rule already contains the right criterion for reactivity control in § 57.30. Elevating the net reactivity coefficient demonstration to an entry criterion under § 57.25 is low-cost, technology-inclusive, directly traceable to the reactivity control safety function, and requires no additional analytical work beyond what § 57.30 already demands. Replace the mass limit with that elevation, and the two entry criteria in proposed § 57.25 will each map cleanly to the safety functions they are meant to represent — one anchored in the NRC's own figure of merit language for radiological consequence, the other anchored in the physics that actually govern reactivity control.

The Broader Connection to the LMP Framework

Practitioners who work across both DOE authorization and NRC licensing space will recognize that this argument is, at its core, an application of the Licensing Modernization Project's underlying principle: regulatory requirements and the analytical work used to satisfy them should be commensurate with the actual safety significance of the function being evaluated, and the metrics selected should map directly to the physical phenomena they represent. I have written about the LMP framework's advantages and limitations in more depth at gibboneynuclear.com.

The SNM mass limit in proposed Part 57 is a step away from that principle — a deterministic threshold embedded in a performance-based framework, applying a physical quantity that correlates imperfectly with the safety function it nominally represents. The good news is that the correction is already written into the rule. It just needs to be moved to the right section.

That window closes June 15.

Sarah Gibboney, P.E. is the founder of Gibboney Nuclear. She has 17 years of continuous nuclear energy experience spanning both DOE authorization and NRC licensing frameworks, has contributed to the licensing of 8 reactor designs and 2 operating reactors, and co-authored 2 construction permit applications. She previously published on the Licensing Modernization Project and NEI 18-04 at gibboneynuclear.com. She works with advanced reactor developers on licensing strategy, pathway selection, and regulatory engagement across both the DOE Reactor Pilot Program and NRC commercial licensing pathways.