MTBF, MTTR and Spares

The reliability and repair statistics that feed every availability model, and how spares provisioning cuts effective repair time.

The two numbers behind availability

Availability rests on two statistics per component: mean time between failures (MTBF, how often it breaks) and mean time to repair (MTTR, how long to fix). A = MTBF / (MTBF + MTTR). Raising MTBF and lowering MTTR are the only two ways to raise availability, and every resiliency technique acts through one of them.

Estimating them honestly

Pre-FOAK, MTBF for many fusion-specific components has no operating history, so it is estimated from component tests, physics-of-failure models, and analog systems, with wide uncertainty. MTTR depends on access, tooling, and spares. We report both as ranges, which is a primary reason the plant availability comes out as the 0.86-0.995 envelope rather than a single figure.

def effective_mttr(diagnose_h, procure_h, repair_h, verify_h, spare_on_hand):
    # a staged spare removes procurement lead time from the repair path
    procure = 0.0 if spare_on_hand else procure_h
    return diagnose_h + procure + repair_h + verify_h

def availability(mtbf_h, mttr_h):
    return mtbf_h / (mtbf_h + mttr_h)

Spares are MTTR insurance

• A staged spare removes procurement lead time - often the largest MTTR term
• Predictive maintenance lets spares be pre-positioned before the failure
• Long-lead items (large coils) dominate spares strategy; some very large items are impractical to spare and are managed by redundancy instead
• Fleet-shared spares pools raise effective coverage across units

For the burner plug coil there is effectively no spare-swap MTTR because a 26.49 T coil is not a shelf item; this is why the plug is treated as a redundancy-and-derate problem, not a spares problem. Spares strategy is an input to availability modeling and scales at the fleet level. This page quantifies only time, never economics, by rule.