Reliability Modeling

System availability is estimated by combining unit reliability, redundancy, and repair time in a fault-tree model.

From one unit to a system

A single unit's availability (0.86–0.995) does not directly give the installation's continuity. That comes from a reliability model that combines unit reliability, the number of redundant units, repair times, and common-cause factors. The model answers the real question: what is the probability the mission loses power over a given period?

What the model captures

A fault-tree or reliability-block model represents how units, cooling, control, and storage combine, and where independence holds or breaks down. It shows that adding an independent unit sharply reduces the chance of a mission outage, while shared components (common cause) can cap the benefit. It also quantifies how storage duration must match repair-and-restart time.

• Combine unit availability with redundancy count
• Model repair time and standby-unit start time
• Represent common-cause couplings explicitly
• Output: probability of mission-level power loss

Honesty in the inputs

A model is only as good as its inputs, and at the design stage the unit failure rate is uncertain (hence the 0.86–0.995 range and the 166–830× regime gap). So the model is used to bound outcomes and to size redundancy conservatively, not to claim a precise system availability. Test-burner data will tighten the inputs.

Reliability modeling here is an engineering discipline whose output is a probability of mission power loss, used to size redundancy conservatively against uncertain design-stage inputs rather than to publish a single confident availability the design cannot yet defend. The model is therefore used conservatively — to size redundancy against pessimistic inputs and to identify which shared components most limit the result — rather than to publish a single confident availability number the design cannot yet support.