Degraded-Mode Power Scheduling
For the burner, deciding what power to promise when a unit is derated - honest firm-power commitments under the availability gate.
Promising only what you can hold
A firm-power customer - a MetroVolt data center, an Aegis fixed defense installation - needs a commitment the plant can keep. Degraded-mode power scheduling decides, given the current health of every unit, how much firm power to promise so that a further single fault does not break the commitment. It is graceful degradation expressed as a dispatch policy, and it is bounded by the honest availability gate.
Firm vs available
Available power is what the fleet can produce right now; firm power is what it can guarantee through the next credible fault. The scheduler commits firm power below available power by a reserve margin sized to the worst single-unit loss, so a failover event causes a reserve draw, not a shortfall.
def firm_power(unit_caps, reserve_units=1):
# firm = total capacity minus the largest 'reserve_units' contributors
caps = sorted(unit_caps, reverse=True)
return sum(caps) - sum(caps[:reserve_units])
# with a stressed plug, a unit's cap is its DERATE point, not its design point
print(firm_power([80, 80, 80, 80], reserve_units=1)) # 240 firm from 320 available
The plug shapes the schedule
- Each burner unit's committed cap is its conservative derate point, because the 26.49 T plug is stressed 3-3.9x at the design bore
- The regime is 166-830x beyond data, so headroom is held back as low-confidence margin
- He-3 fuel availability (~400x supply gap) is a scheduling constraint, not just a design one
- Facility firm power leans on N+k units, never a single machine, per the 0.99982 gate
The scheduler never promises past the availability envelope; if health degrades, the firm commitment steps down transparently rather than being missed. It draws unit health from predictive maintenance and reserve sizing from fleet availability. Nothing here is expressed in economic terms - only megawatts and time.