Loading site…
Skip to main content
Abinauv Selvaraj logoAbinauv Selvaraj
All Projects
Operations Research

Optimization of E-Commerce Packaging Options

Master's dissertation project for Lakeland (UK) — a Mixed Integer Programming model that selects the optimal set of box sizes for e-commerce fulfilment, minimizing wasted space and packaging material across 40,000+ historical orders.

Confidential — Source code not publicly available

About This Project

Developed during an Operations Research internship at Lakeland Limited (UK) as a Master's dissertation at Lancaster University. The system uses a two-stage Mixed Integer Programming (MIP) approach: first, a 3D bin packing model (Padberg formulation) determines which candidate boxes can physically fit each order's products using orthogonal placement and non-intersection constraints. Then, a box selection MIP chooses the optimal subset of box types that maximizes packing efficiency or minimizes total volume across all orders. A stochastic simulation framework using sample-average approximation validates generalizability across 20 trials with in-sample sizes of 500–10,000 orders. The project demonstrated significant potential for reducing packaging waste and shipping costs in e-commerce operations.

Tech Stack

PythonGurobiPuLPPandasMixed Integer Programming

Key Features

Two-stage MIP: 3D bin packing + box selection optimization
Padberg formulation for orthogonal 3D packing constraints
Stochastic simulation with sample-average approximation
Evaluated on 40,000+ real e-commerce orders from Lakeland
Dual objectives: maximize packing efficiency or minimize total volume

Frequently Asked Questions

What is Optimization of E-Commerce Packaging Options?▾
Developed during an Operations Research internship at Lakeland Limited (UK) as a Master's dissertation at Lancaster University. The system uses a two-stage Mixed Integer Programming (MIP) approach: first, a 3D bin packing model (Padberg formulation) determines which candidate boxes can physically fit each order's products using orthogonal placement and non-intersection constraints. Then, a box selection MIP chooses the optimal subset of box types that maximizes packing efficiency or minimizes total volume across all orders. A stochastic simulation framework using sample-average approximation validates generalizability across 20 trials with in-sample sizes of 500–10,000 orders. The project demonstrated significant potential for reducing packaging waste and shipping costs in e-commerce operations.
What technology stack does Optimization of E-Commerce Packaging Options use?▾
Optimization of E-Commerce Packaging Options is built with Python, Gurobi, PuLP, Pandas, Mixed Integer Programming.
Is Optimization of E-Commerce Packaging Options open source?▾
No, the source code for Optimization of E-Commerce Packaging Options is confidential as it was developed for a commercial client. However, the methodology and approach are described on this page.

Need something similar built?

I build custom tools and systems tailored to your business requirements.

Get in Touch