Port Central
Brain
A centralized intelligence platform driven by one shared global objective and three coordinated optimization layers that minimize weighted vessel sojourn time across the entire port.
Foundational Architecture
Single Source of Truth
Ontology as World ModelA unified, real-time graph that represents every vessel, crane, AGV, yard slot, and container in the port. No silos, no stale data. One model, one truth.
Globally Aligned Optimization
One Objective, Three LayersOne shared global objective — minimize weighted vessel sojourn time — decomposed into three MECE time-axis layers. Each layer solves its own subproblem, but never for its own sake. Always aligned with the global objective.
Bidirectional Coordination
Iterative Cross-Layer FeedbackLayers negotiate, not just cascade. The Service layer sends infeasibility cuts, congestion penalties, and fragility scores back to Pre-Berth. Neither isolated local optimization nor monolithic single-shot solving.
Event-Driven Intelligence
Rolling-Horizon ReoptimizationThe system doesn't wait for full replans. Rolling-horizon reoptimization and trigger-based replanning detect deviations as they emerge and adjust continuously under changing conditions.
Five Port Layers
Container flow moves through five physical layers, from sea to gate. These map to three optimization layers along the vessel sojourn time axis.
Covers Sea Interface. Vessel admission sequencing, berth assignment, and start timing. Minimizes Tberthstart - Tready.
Covers Vertical Transfer + Horizontal Transport + Storage. STS cranes, AGV dispatch, and yard-side absorption. Minimizes Tworkdone - Tberthstart.
Covers Gate Complex for departure. Departure clearance, berth release, and spillback prevention. Minimizes Tdep - Tworkdone.
Service sends infeasibility cuts and congestion penalties back to Pre-Berth. Layers negotiate iteratively, not cascade one-way.
Explore the System
Ontology Structure
17 node types, 26 relationship types. The complete graph schema that models every entity and connection in port operations.
Optimization Engine
Three MECE time-decomposition layers coordinated under one global objective. Hierarchical decomposition with bidirectional feedback.
Pre-Berth Layer
Minimize admission delay. Berth assignment, start timing, and vessel admission sequencing before service begins.
Service Layer
Minimize berth service time. STS crane sequencing, AGV dispatch, and yard-side absorption. The most operationally critical layer.
Post-Service Layer
Minimize release delay. Departure clearance, berth release, and spillback prevention after service completion.