Python and Gurobi: Bakach et al. (2019), A Two-Tier Urban Delivery Network

Modeling the approach of "Bakach et al. (2019), A Two-Tier Urban Delivery Network" in Python using Gurob 9.0.1

Setting up the working environment

• install Anaconda/Miniconda, Python and Gurobi following https://www.gurobi.com/documentation/quickstart.html (including obtaining and activating a gurobi license)
• open the conda command prompt, navigate to this repository's location and create a new conda environment. You can use the environment.yml file so some basic necessary packages (e.g. for the examples) will already be installed with it. (Ensure that the file is UTF-8 encoded, if problems occur)

conda env create -f environment.yml

• activate the the environment by running

conda activate gurobi

• if you have not used the environment.yml file, install the gurobipy package by running the following command in the conda shell

conda install gurobi

• as this repo is based on jupyter notebooks, run

jupyter notebook

to open the jupyter interface

Examples

• to get familiar with the use of Gurobi and Python, follow the examples at https://www.gurobi.com/documentation/8.1/quickstart_windows/py_simple_python_example.html

• some of the examples can be found in the corresponding notebooks in the examples folder. However, they are not exactly as they are presented in the website guide but merely a few simple test snippets to get used to the gurobi modeling procedure

Modeling the Two-Tier Urban Delivery Problem

• The models without time windows can be found in the TTUD folder
• Each of the notebooks(with and without time windows) has two models: ** Tier 1: minimizes the number of necessary hubs ** Tier 2: optimizes the necessary time by using only the minimum number of hubs as determined by tier 1
• More details are given in the notebook (markdown cells and comments)

Notice

• Most of the code is written with focus on readability rather than efficiency
• The interactive plots require a live ipython session (i.e. they are only interactive when they are executed in a notebook)