VALAR WATCH
Key findingsTheir claims, annotatedEvidenceTake action
374 facts · 512 sources · updated 2026-09-30

Fact safe-020

newcompany claimreported 'Walk-away' cooling: 72 hours without operators in Valar's safety agreement, 'indefinitely' in DOE's review
2026-06-26 · safe-020 · Valar Atomics, U.S. Department of Energy

Valar's NSDA sets target performance for decay-heat removal: capacity for 100% of decay heat, natural circulation within 30 minutes, and 'No operator action required for 72 hours' (page 23); it credits a reactor cavity cooling system, a vessel auxiliary cooling path and conduction to the structures. DOE's categorical exclusion (April 2026), describing the project's attributes, states that the passive system keeps fuel below safety limits 'indefinitely' without operator action or outside power. Valar's Castle Country page says that on 26 June 2026 it shut Ward 250 down, switched off every active cooling system, simulated a total loss of electrical power and watched the reactor cool itself. That test came eight days after the first criticality DOE announced (18 June). ANS reported Valar's announcement that it reached 10 kWt on 22 June, and Valar's Castle Country page says full power came 'days later' than criticality; no DOE or independent record of the test, and no power log before it, was found; how much heat a reactor must shed after shutdown depends on how long and how hard it ran, which the NSDA's own accident method ties to operating history (safe-009). The claim of passive cooling is consistent with the design documents; the June 26 demonstration rests on the company's account.

In plain termsAfter a reactor shuts down, its fuel keeps making heat for a while, and that heat has to go somewhere. Valar's design is meant to shed it through air flow and the surrounding structures with no pumps and no people. Its own safety agreement promises this for 72 hours; DOE's environmental review states a stronger 'indefinitely'. The company says it proved it in a public test on June 26, 2026, but no outside record of that test, or of how much the reactor had run beforehand beyond the company's own announcements, has been published.
verified 2026-09-29: Re-read NSDA pp. 14-15, 23, CX p. 5, Castle Country and ANS on 2026-09-29. Holds as a company claim (one company account). Fixed: 'no published power history' was too strong; ANS relayed a 22 June 10 kWt announcement and Valar claims full power 'days later'. 'Repeated' changed to 'states' (nothing shows DOE took the wording from Valar). Correction 2026-09-29 (wording review): wording restated.

In tabs: safety, operations · file safety