Uncertainty-Aware Digital Twin Framework for Resilient Low-Carbon Port Infrastructure under Electrification and Sensor Degradation

Abstract

The electrification of port operations is reshaping the technical relationship between logistics performance, energy demand, infrastructure condition and data reliability. Although digital twins are increasingly used for port monitoring and decision support, many frameworks continue to assume clean sensing, synchronised data streams and stable observability. This study develops an uncertainty-aware digital twin framework for resilient low-carbon port infrastructure under electrification and sensor degradation. A synthetic reference container terminal with 4.5 million TEU/year throughput, six berths, twenty-two quay cranes, 58,000 TEU yard capacity, 120 monitoring channels, 55 MW grid-import capacity, 28 MW renewable capacity and 60 MWh battery storage is simulated for one year at a 5 min update interval. Fifty paired stochastic replications are analysed after a 14-day warm-up period. Under compound stress comprising 35% sensor loss, 5% measurement noise, 10 min synchronisation delay, 90% electrification, 25% higher latent vessel-arrival demand intensity before application of the approach-channel cap, 25% lower renewable availability and 12% lower equipment productivity, the framework yields 10.5 h mean completed-vessel waiting delay, 5.7% energy shortfall, 0.27 infrastructure-health risk, 0.31 uncertainty index, 14.6 h recovery time and a resilience index of 0.64. Relative to a conventional digital twin, completed-vessel waiting delay decreases by 18.0%, energy shortfall by 32.1%, uncertainty index by 29.5% and recovery time by 26.3%. Paired replication-level tests remain significant after Holm adjustment, with large paired effect sizes for the principal operational, energy, uncertainty and resilience outcomes. The results show that resilient electrified-port operation depends not only on installed electrical capacity, but also on temporal alignment, probabilistic state estimation, energy flexibility, health-aware operation and explicit representation of degraded information quality.