Case Study - Optimizing delivery operations through decision analytics

Overview

E-Bebek's delivery operations faced a common but difficult challenge: coordinating a mixed fleet across hundreds of daily deliveries with strict time windows, capacity constraints, and varying service requirements. Route planning relied heavily on manual decisions and dispatcher experience, resulting in suboptimal routes and inconsistent cost structures.

yilven approached this as a decision analytics problem — not simply a routing task. The goal was to build a system that models the full operational reality: vehicle types, capacity limits, time windows, service durations, and cost structures — and produces optimized plans that account for all of these constraints simultaneously.

What we built

The core of the system is a Mixed-Integer Linear Programming (MILP) model formulated for the Operational Vehicle Routing Problem with Time Windows (OVRPTW). The model was implemented using Gurobi, enabling exact optimization for scenarios where computational time allows.

To handle larger-scale instances and provide faster approximate solutions, a Genetic Algorithm heuristic was also developed in Python, designed to explore the solution space efficiently while respecting all operational constraints.

The system functions as a decision support platform — providing optimized route plans, cost projections, and scenario comparisons that allow operations teams to make informed planning decisions rather than relying on intuition alone.

What we delivered