Data for decarbonization: Emissions and flexibility findings from data-driven thermal generator analysis

The electric grid is transforming more quickly than ever before, from centralized to decentralized, dispatchable to variable, and fossil to renewable. And yet climate change necessitates even faster decarbonization. This thesis formulates data-driven methods for analyzing, improving, and decarbonizing the electric grid. Methods are applied to the Western Interconnection (WI), but can be used in any electricity system.

First, major trends in emissions and operations are visualized and contextualized with policy and market background. Market-based policy and cheap natural gas drove down SO2 and NOx emissions over the last two decades. CO2 is not priced, except in California, and has see weak reductions over the same period. Second, data-driven methods are developed to find minimum load (Pmin), maximum load (Pmax), and maximum ramp rates (Ru, Rd) for thermal generating units (TGU). These parameters explain TGU flexibility, which is critical to renewable integration, Comparison to standard values reveals substantially more existing TGU flexibility, in the form of lower minimum load and proven ramp rates. This finding uncovers an extremely rapid and low-cost way to add power system flexibility: incorporating data-driven TGU parameters into grid operation models. Third, the impact of variable renewable electricity (VRE) on carbon intensity (CI) of TGUs is quantitatively assessed. The impact so far has been small, but impacts could increase if TGUs spent significant amounts of time operating at low load factors. Findings suggest that our understanding of the grid could be improved through novel data-driven analysis of emissions trends, operations trends, TGU parameters, and system impacts of VRE. These analyses clarify a path toward decarbonized electricity.