Post-disaster demand persisting past the surge window

System failure identification

Demand remains elevated for weeks after an incident, driven by deferred care and follow-up. Surge structures stand down on a schedule set by the incident rather than by the demand. Capacity is withdrawn while the demand it was created for is still present.
Data pipeline diagnostics

Escalation and de-escalation are both keyed to the event. Nothing models the tail, so the recovery period is planned as though demand returns to baseline when the incident closes. The tail is the longest part of the episode and the least modelled.
Predictive computational model

Forecast the decay curve from comparable prior incidents and set de-escalation against observed demand rather than elapsed time since the event. Prior incidents supply the shape of the curve even where magnitude differs.
Workflow integration

Write the criteria into the de-escalation section of the plan itself. Standing down is a decision made under pressure to return to normal, so it needs a written trigger. An unwritten trigger loses to organisational pressure every time.

System failure identification