Specialist / long-horizon capabilities (backlog)

[derrynknife]

Tracking issue for the longer-horizon capabilities. Split each into a dedicated issue as it is prioritised:

• Phased-mission systems — the diagram/config changes per mission phase.
• Multi-state / capacity (flow) systems — degraded/derated states and throughput, not just working/failed.
• Undirected network (s-t) reliability — reliability on general graphs (power/pipe networks).
• Markov / state-space models for dependent repairable systems (shared repair crews, switching failures) beyond simulation.
• Reliability growth (Crow-AMSAA / NHPP, Duane) & reliability demonstration test planning.
• RBD-level maintenance & spares optimisation (RCM layer). A layer on top of RepairableRBD: given structure, per-component cost/duration and the component economics (NonRepairable/Repairable, from Decide the fate of the vestigial Repairable class #33), choose per-component policies to meet a system availability target at minimum cost — accounting for redundancy, which is why this is system-level and not per-component. Covers block replacement, inspection intervals, condition-based policies (ties to Condition-based ("digital twin") reliability evaluation layer #36) and spares forecasting / stocking (fleet aggregation via the renewal function). This is the deliberate home for the availability-optimal / block-replacement / spares items kept out of the component classes.
• BDD/ZBDD engine for large-system scaling.

Estimate

Effort: — (umbrella; each item is L–XL on its own) · Value: Varies

Triage-only as an issue — it's a holding pen, so no single estimate applies. Individually some items are high value when their moment comes (BDD/ZBDD for large-system scale; Markov for dependent repairable systems), others are specialist. Promote an item to its own estimated issue when it's prioritised.