Transport infrastructure performance under repeated earthquakes in Kumamoto

Cities routinely rebuild critical transport infrastructure after disasters, but completed projects rarely face a second documented shock that can show whether structural and network-capacity adaptations remain operationally credible under renewed stress. We develop a bounded repeated-shock design that links official intervention chronologies, event-specific shaking, and post-event performance, with an exhaustive denominator of 65 candidate records from 4 official universes and separate outcomes for the two retained cases. At Yokoe Bridge, a matched-endpoint analysis shows that the same asset fully reopens within 19.61 days after the later earthquake, compared with 1022.4 days after the earlier earthquake, and every admissible restoration-time ratio remains below one; estimated MMI, PGA, and pseudo-spectral acceleration at three common periods are higher in the later event. At Mashiki Central Line, corridor passage is observed in 20 of 21 available post-shock snapshots and near-full longitudinal coverage appears no later than 43.55 h, although the earlier restriction records and later probe traces remain non-equivalent outcomes. These asymmetric cases show how cities can use repeated hazards to audit adaptation investments when asset identity, chronology, measurement, and claim boundaries remain explicit, supporting pathway-specific infrastructure learning without implying a causal reconstruction effect or a population-wide estimate.