Gaurav Khalegaonkar

Project title: Designing a novel method based on machine learning to analyze the results of MGA-based energy system optimization models

The energy sector has hundreds of technologies and millions of actors working together to balance the system. Analyzing the impact of changes in one or more components on the entire system is a challenging task. Researchers and policymakers are using computer-based energy system models to understand these techno-economic impacts. Usually, the energy system model generates a single optimized scenario of a future energy system. However, a single scenario is not always beneficial due to uncertainties involved in the basic assumption of the energy system model. Energy system modellers started to integrate uncertainties in their model and generate multiple near-optimal scenarios. Model to generate alternative (MGA) is one such type of method which generates hundreds of equally possible future energy system configurations. Analyzing hundreds of configurations with multiple technologies at multiple locations is a challenging task. In this thesis, I designed a method and built a Python package based on the designed method to facilitate smooth analysis of MGA-based energy system optimization model results. The designed package cluster the solution space identifies a suitable number of clusters given technology and gives a representative solution for each cluster. This allows smoother analysis as compared to directly going through the results. Along with that application MGA analysis package is demonstrated on the results of the sector-coupled euro calliope model (Pickering et al. doi.org/jbd7). The MGA analysis package will be made available on the Git-hub page for application after the defence of the thesis.

What is the contribution to the Energy Transition Lab?
This project directly contributes to one of the pillars of the energy transition lab, which is energy system modelling. The output of this thesis project is an interactive python based dashboard which can be used
to analyze the solution of MGA based energy system optimization model. The MGA analysis package will encourage modellers to use the MGA analysis method as analysis of results will become easier with the help of the package.