Learning by modeling energy systems

Abstract: Meeting the 2°C climate target would likely require reducing carbon dioxide emissions from the global energy system to virtually zero within 50-100 years, and within 30-50 years for the 1.5°C target. Both cases would involve a complete transition of the global energy system to zero-emission technologies like renewables or nuclear power at unprecedented rates. This complex challenge can only be analyzed with energy system models, i.e. large computer models that can generate future energy scenarios. This thesis presents five papers that develop methodology for modeling the global energy transition. In papers 1-2, we develop new methods for representing technological development of emerging technologies like solar or wind power in energy models. We use “experience curves”, empirical relationships that describe how costs tend to fall for new technologies as a function of their market growth. We find that by investing in solar and wind at a global scale we can drive down costs to a point where they compete with conventional fossil energy sources. Paper 3 is a study of meeting climate targets with bioenergy with carbon capture and storage (BECCS) using an integrated energy-climate model. BECCS is a technology that can produce negative emissions; i.e., it can deliver energy while actively removing CO2 from the atmosphere. We find that if BECCS is used on a global scale, it can significantly reduce costs of meeting the 1.5°C target and potentially reverse global warming in the long run. Paper 4 addresses another modeling problem. Many global energy models are too large to use an hourly time resolution which may be necessary to represent very high penetration levels of variable renewables like solar and wind power. We present a method called “resource-based slicing” that can capture sufficient variability in just 16 annual time periods. Finally, in paper 5, we develop an open-source code base that uses global meteorological datasets to generate all input data an energy model needs to study solar-, wind- and hydropower in arbitrary world regions. Our GIS-based approach produces both hourly capacity factors and regional potentials for installed capacity, and our simple generic model performs on par with more detailed dedicated models of European electricity generation.