Supply Chain Network Control Tower

Compare recovery policies and see where a port disruption moves the bottleneck.

Industry: Logistics & distribution
Business context: A fictional logistics network moving containers through a port and a distribution center.
Data sources: Simulated demand, capacity, and disruptions used to compare recovery policies.

When a port loses capacity, faster unloading does not necessarily restore the network. This experiment follows each container through a port and distribution center to compare five operational responses.

Observe → Understand

Track daily throughput, queues and original delivery commitments. A three-day port disruption creates a backlog; increasing port capacity can move that backlog downstream.

Decide → Execute

Compare no intervention, extra port capacity, extra distribution capacity and coordinated recovery against normal operations. Python runs 500 paired simulations using identical demand and capacity shocks for every option.

Measure

Inspect missed commitments, peak distribution backlog and queue recovery across P10, median and P90 outcomes. Trace individual containers without changing their original due dates.

Evidence and limitations

EYESTERION is fictional. Operations are synthetic and the parameters are not calibrated to a real business. This is a service trade-off analysis, not a cost optimizer or order-level OTIF model. A clearly labelled reference snapshot keeps the demo usable when its live service is unavailable.

Architecture

Python/FastAPI on OVH · n8n orchestration · private Supabase schema · independent protected Vercel dashboard · server-side portfolio gateway.