Group assignment 2.5: Waterschap ‘dam

Group assignment 2.5: Waterschap ‘dam#

Due: Friday, 12th December, 2025.

You can access this assignment with the following link: https://classroom.github.com/a/lKG__wqh

Note: You will need to install pymoo to run the notebook in Part 2. This is a different package from pyomo. So, not a typo!

The notebooks contain a lot of cells and code, but a lot of it has been implemented for you beforehand. Please focus on reading the problem parameters carefully and implement the given constraints in the relevant code cells. For the notebook in Part 2, you do not need to write any code, only understand what is happening and adjust parameters.

This GA is in two parts:

Part 1: MILP

In this assignment you will implement a Mixed-Integer Linear Programming model in Pyomo + HiGHS for the Amsterdam HoReCa waterborne logistics network design problem. You will:

  • Work with a realistic dataset

  • Implement the MILP model presented in the book chapter

  • Explore how the solution changes when emission penalties are varied (via \(\lambda\))

You are not asked to derive the mathematical formulation from scratch here – it is already given in the book. Your task is to translate it into code and experiment with it.

Part 2: GA (Genetic Algorithm)

Note: part of the background material for this project was already available in Chapter 5.11 of the textbook.

In this notebook, we illustrate how a metaheuristic approach, specifically a genetic algorithm (GA), can be used to obtain high-quality (though not guaranteed optimal) solutions more efficiently.

We showed in the previous notebook how to use MILP to solve the Waterborne Urban Logistics Problem in Amsterdam. As you observed, computation times increase rapidly as the problem grows in scale. For the full Amsterdam case—including all HoReCa entities—the MILP solver cannot reach optimality within reasonable time.

Because of time constraints during the Friday session, we continue working with the same problem size as before (300 demand zones and 36 TPs). For this scale, the performance benefits of a GA may be limited. Nonetheless, metaheuristics become especially valuable for larger instances, where exact methods struggle (e.g., For 1500 HoReCa entities, GA obtained near-optimal solutions in less than 10 minutes, while MILP was not able to reach the optimal solution within an hour.)

You can preview this assignment on https://mude.citg.tudelft.nl/workbook-2025/assignments/GA2.5/README.html. After the deadline, this link will include solutions. The preview of the assignment version is shared here: https://mude.citg.tudelft.nl/workbook-2025/no_solutions/assignments/

By Nadia Pourmohammadzia and Gonçalo Homen de Almeida Correia, Delft University of Technology. CC BY 4.0, more info on the Credits page of Workbook