Truck & Dock Optimization

The problem

Yard and dock operations look simple from the outside: trucks show up, get assigned a dock, get loaded or unloaded, and leave. In practice, arrivals are irregular, service times vary by load type and equipment, and not every dock can handle every truck. Small changes in how assignments are made can swing congestion and idle time significantly — but that's hard to see without a model of the system in motion.

The goal of this project was to build that model: a simulation engine realistic enough to test dock-assignment strategies before trying them on a live yard.

How it's built

• Modeled truck arrivals using realistic probability distributions rather than fixed intervals, so the simulation reflects the bursty, uneven way trucks actually show up.
• Layered in service-time variability, since loading/unloading time depends heavily on load type, crew, and equipment.
• Added real operational constraints: equipment type compatibility, dock capability, and priority loads that need to jump the queue.
• Once the simulation was stable, experimented with different dock-assignment strategies to see how assignment logic — not just capacity — drives congestion and idle time.
• Used AI to speed up the early phases: generating scenario variations, drafting quick heuristics, and visualizing throughput patterns — while grounding the actual model in real operational behavior.

Try the underlying model

This is a simplified version of the core mechanics: trucks arrive at random (a Poisson process), each dock serves one truck at a time with variable service duration, and a truck waits for the next free dock. Adjust the inputs and watch throughput, wait time, and dock utilization respond — the same dynamic the full simulation explores at much higher fidelity.

Where it's heading