E-Commerce Warehouse Optimizer
Supply chain optimization using linear programming to minimize warehouse costs and improve distribution efficiency.
An e-commerce company with 6 warehouses, 10 delivery regions, and 300+ products was running at 45% order fulfillment — more than half of all orders resulted in stockouts. The root cause: inventory was allocated by gut feel, not math. This project formulates the allocation problem as a linear program and solves it to optimality.
The optimization model minimizes total system cost (transportation + holding + stockout penalties) subject to capacity, demand, and service-level constraints. The key insight was modeling inventory as continuous flow capacity rather than static stock — a 50,000-unit warehouse turning over 12× per year has 600,000 units of annual flow capacity, not 50,000.
The result: fulfillment jumped from 45% to 99%+, stockouts dropped below 5%, and annual profit improved by $44.9M. The system also identified a $3.7M transportation cost savings opportunity through carrier renegotiation. Nine scenario analyses (capacity expansion, cost changes, higher service targets) provide a decision playbook for the operations team.
At 100% capacity with 12× annual turnover, the network achieves 100.0% fulfillment at $2118K total cost. All regions are served at near-perfect fulfillment. Transport costs account for $734K (35% of total).
- ▸Multi-commodity network flow LP: decision variables x[i,j,p] for shipments from warehouse i to region j for product p, with stockout slack variables.
- ▸Objective: minimize Σ transport_cost × shipments + Σ holding_cost × inventory + Σ penalty × stockouts. Solved with CBC (branch-and-cut) in ~2 seconds.
- ▸Scenario engine runs 9 variants (capacity ±10-30%, transport cost ±10-20%, service target 95-99%) and compares KPIs in a unified dashboard.