Improving the simulation of variable renewable energy in the MEDEAS integrated assessment model

Abstract

This master thesis explores the application of a surrogate model to achieve the integration of electricity network constraints into integrated assessment models, that attempt to predict the evolution of the main features of the society, including economic, demographic, climatic and energy factors. Tools were developped to run Dispa-SET, a dispatch model for the European electricity network, to build a dataset of simulations on adequate sample points. This dataset is then used to train the surrogate model employing appropriate machine learning methods. Finally, this model is integrated into the MEDEAS IAM, connecting their variables in a meaningful way, and the resulting model is run for different scenarios. Results show that integrating the model in MEDEAS leads to a lower prediction of the production of variable renewable energy sources compared to the initial MEDEAS output.