Passenger Rerouting Engine

Overview

Passenger Rerouting Engine is a proof-of-concept recovery simulator for airline disruption management. It starts with a delayed inbound feeder, identifies passenger groups whose onward connections are missed or statistically fragile, and asks a network-wide optimizer to find feasible alternatives.

Unlike a simple shortest-path tool, the engine treats every remaining seat as a shared constraint. Passenger groups compete for the same available inventory, so the recommendation can split traffic across multiple routings when one attractive option runs out of seats.

The frontend is hosted as an AviationDecoded project, while the optimization logic runs in a Python backend using Google OR-Tools.

Why I Built It

Disruption recovery is where airline networks become painfully real. A delay is not just a late aircraft; it creates misconnected passengers, baggage risk, seat shortages, customer priority decisions, and downstream operational cost.

I built this prototype to make that logic visible. The goal is not to mimic a production rebooking system, but to show the underlying optimization problem: every itinerary is a path through a time-expanded network, and every available seat is a capacity constraint.

What It Demonstrates

• Time-expanded flight graphs
• Minimum connection time as a hard feasibility rule
• Misconnection prediction rate as a soft risk cost
• Capacity-aware passenger group assignment
• Multi-hub recovery across Frankfurt, Munich, Zurich, Vienna, Brussels, and Rome
• Partner-hub fallbacks via Amsterdam, Paris, London, and Madrid
• Escalation logic when no automated answer is good enough
• Why the best path is not always the shortest or earliest path