Computing for Tritium Fuel-Cycle Accounting

Tracking every gram of tritium through breeding, extraction, storage, and burn so the fuel cycle closes on paper before it must close in hardware.

Why accounting is hard

Tritium is bred in the blanket, extracted, stored, injected, partly burned, and partly recovered from exhaust. It also decays (12.3-year half-life) and permeates through hot metal walls. Closing the cycle means the plant breeds at least as much as it consumes plus losses. Getting this right is a bookkeeping problem with physics inside every entry.

The mass balance

At steady state, tritium produced must cover tritium burned, tritium lost to decay, and tritium lost to permeation and holdup, with a working inventory buffering the flows. The breeding ratio must exceed one with margin because real extraction and recovery are imperfect. The Hyperion breeder targets a tritium breeding ratio near 1.8 to carry that margin.

• Sources: neutron capture in the breeder blanket (lithium)
• Sinks: fusion burn, radioactive decay, permeation losses, processing holdup
• Inventories: storage beds, plasma-facing holdup, processing lines

def tbr_required(burn, decay, perm, recovery_eff):
    losses = burn + decay + perm
    return losses / recovery_eff   # bred tritium needed per burned unit

print(round(tbr_required(1.0, 0.05, 0.08, 0.9), 3))

What computing adds

Neutronics codes estimate breeding; permeation and decay models estimate losses; a system model ties them into a time-dependent inventory. Uncertainty is propagated so the answer is a band, not a point. The goal is to demonstrate cycle closure in simulation with honest error bars well before first-of-a-kind first tritium around 2030.

Honesty gate

Local breeding in an idealized geometry differs from net breeding in the real machine with ports, gaps, and structure. Accounting keeps those distinct so a favorable local number is never mistaken for a closed net cycle.