A Heterogeneous Fleet Approach to the Capacitated Multi-Depot Vehicle Routing Problem with Time Windows for Indian Cooperative Dairy

This repository contains the dataset preprocessing pipeline, optimization solver code, and analytical results for the paper submitted to the 11th International Conference on Soft Computing: Theories and Applications (SoCTA 2026) at SVNIT Surat.

🚀 Overview

Milk collection in Indian cooperative dairies is a critical logistical challenge due to product perishability, geographically dispersed collection centers, and strict temporal constraints. This project implements a Capacitated Multi-Depot Vehicle Routing Problem with Time Windows (CVRPTW) supporting a heterogeneous vehicle fleet.

Key Contributions:

• Real-World Road Geometry: Integrates Google OR-Tools with the Open Source Routing Machine (OSRM) to eliminate Euclidean distance errors, which typically distort Indian rural road logistics by 30-50%.
• Heterogeneous Fleet Optimization: Models 378 collection points using a heterogeneous fleet tailored to Indian rural constraints, reducing daily operational costs by 60.4% compared to a homogeneous heavy-tanker fleet.
• Strict Perishability Constraints: Enforces a 4-hour maximum shift duration using time-dimension constraints and Guided Local Search (GLS) with node disjunctions.

📁 Repository Structure

├── data/                       # Raw and processed optimization datasets (JSON/CSV)
├── graphs/                     # Generated evaluation plots and figures
├── results/                    # Optimization outputs, benchmark logs, and sensitivity analysis
├── src/
│   ├── data_prep/              # OSRM matrix generation and data cleaning scripts
│   └── vrp_ortools/            # Core OR-Tools Constraint Programming models
├── socta_dairy_mdvrp_source/   # LaTeX source code for the SoCTA conference paper
├── requirements.txt            # Project dependencies
└── README.md

🚦 Getting Started

1. Install Dependencies

Ensure you have Python 3.8+ installed, then run:

pip install -r requirements.txt

2. Generate Distance Matrix

Run the data preparation script to generate the real-road distance matrix via OSRM API:

python src/data_prep/prepare_global_data.py

3. Run Optimization

Execute the core OR-Tools solver to generate optimized routes across all depots:

python src/vrp_ortools/solve_mdvrp.py

📄 License & Citation

The code in this repository is available for academic and research purposes. If you build upon this work, please cite our SoCTA 2026 paper appropriately.