Markov Chain Model Integrates Seismic Reliability and Recovery Dynamics

The preprint proposes embedding recovery states directly into system dynamics rather than treating them as post-processing steps, replacing Poissonian assumptions with state-dependent transitions. Transient analysis yields time-dependent failure probabilities while the expected operational time fraction quantifies resilience, computed via spectral decomposition of the generator matrix. Results indicate recovery rates alter long-term resilience metrics even when conventional reliability indices remain stable.

Standard PBEE frameworks developed for single-event assessment now face pressure from compound seismic sequences and aging infrastructure under climate-amplified hazards. The Markov approach connects these by tracking cumulative damage without memoryless renewal, addressing gaps in life-cycle analysis for critical facilities where downtime costs dominate.

Prior work on Markov models in reliability engineering and FEMA P-58 loss assessment treated recovery separately; this integration reveals sensitivity differences that isolated reliability calculations miss. Explicit recovery modeling becomes essential when repeated loading intervals shorten due to regional seismicity increases.

Next steps require calibration against empirical recovery datasets from recent earthquakes and extension to portfolios with spatially correlated ground motions to test scalability beyond the two archetypes examined.