A Performance-Based Earthquake Engineering Method to Estimate Downtime in Critical Facilities

A method is offered to estimate downtime (the time to restore a facility to operability, not total repair time) for second-generation performance-based earthquake engineering. It applies in cases where downtime is primarily controlled by nonstructural systems (although red-tagging and lifeline failure are addressed); restoration begins immediately after the earthquake; and components are repaired in parallel. It uses fault trees to relate component damage to a building’s post-earthquake operability; component fragility functions to estimate the occurrence probability of basic events; and uncertain component repair times conditioned on various possible damage states. The method meets four criteria for second-generation performance-based earthquake engineering: (1) it estimates a system-level performance metric of interest to the facility stakeholder, i.e., time required to restore some critical function that defines the top event; (2) it acknowledges and propagates all important sources of uncertainty; (3) it (can) employ nonlinear dynamic structural analysis; and (4) it considers the detailed components that comprise the system. The methodology is illustrated using a real facility, in particular a computer data center. Results compare reasonably with HAZUS-MH, ATC-13, the actual downtime history of the facility, and the judgment of an operator of the facility who participated in the development and review of the fault tree.