Hybrid mini-grid optimization

This repository contains the code to perform an optimization of the generation system of a PV/diesel/hybrid mini-grid system. A Particle Swarm Optimization (PSO) algorithm is applied to search the solution-space and identify the combination of PV capacity, diesel generator capacity and battery capacity that can meet the demand at the lowest cost. The initial methodology is outlined in this MSc thesis, and further improved with the PSO algorithm as described in this pre-print.

The optimization is done by simulating the system over each hour during one year. To do so, hourly PV resource availability and temperature data is required. A module is developed that automatically retrieves such data from renewables.ninja. To do so, the user needs to provide an access token to the renewables.ninja API - see instructions here.

Content

This repository contains:

• The source code of the mini-grid simulation algorithm and PSO implementation
• An environment .yml file needed for creating the python environment to run the scripts using Anaconda.
• Two example notebooks. The first provides an example implementation that optimizes a single settlement, and the second an example implementation that automatically optimizes multiple settlements through a case of Sierra Leone.

Installing and running the clustering notebook

Requirements

The scripts in this repo have been developed in Python 3. We recommend installing Anaconda's free distribution as suited for your operating system.

Install the clustering repository from GitHub

After installing Anaconda you can download the repository directly or clone it to your designated local directory using:

> conda install git
> git clone https://github.com/AndreasSahlberg/hybrid-mini-grid-optimization.git

Once installed, open anaconda prompt and move to your local directory using:

> cd ..\hybrid-mini-grid-optimization

In order to be able to run the tool you have to install all necessary packages. The "environment.yml" files contains all of these and can be easily set up by creating a new virtual environment using:

conda env create --name mgopt --file environment.yml

This might take some time. When complete, activate the virtual environment using:

conda activate mgopt

With the environment activated, you can start by testing the example notebooks in a "jupyter notebook" session by simply typing:

..\hybrid-mini-grid-optimization > jupyter notebook