UC Berkeley

Global warming is one of the most significant problems facing humanity, and reducing emissions from the electricity sector is critical for mitigating global warming impacts. My work here focuses on developing computational tools to plan cost effective mitigation pathways for the electricity sector and using them collaboratively. The complexity and scale of globally transitioning electrical power grids away from fossil fuels over the coming decades will require a large-scale collaborative effort with effective coordination of many actors trained in diverse disciplines. Historically, energy-modeling efforts have tended to be siloed and fragmented between and even within research groups. In my research I have attempted to provide an alternative to that status quo by improving an open source renewable planning model, Switch, increasing its usability and accessibility to interdisciplinary researchers, and collaboratively applying it to mitigation planning.

We used the Switch model to conduct detailed research into cost effective mitigation pathways for the Western portion of North America, or the WECC power grid. We found that renewable portfolio standards were insufficient to meet climate stabilization goals, and more targeted policies were needed that specifically focused on emission reductions. We identified investment plans that could lead to dramatic decreases in emissions without significantly increasing electricity costs over the next twenty years by retiring coal and replacing it with natural gas and renewables while evolving the grid to better accommodate variable renewable energy.

We found that meeting overall 2050 targets will require concerted action on many fronts, including aggressive efficiency programs, electrification of transportation and heating, and dramatically reducing emissions from the electricity sector. Meeting 2050 emission goals without significantly increasing energy costs also will require additional technological innovation. Two promising technological pathways for long-term cost containment are developing low cost solar in conjunction with low cost storage or demand response, and developing Biomass Energy with Carbon Capture and Sequestration (BECCS) to provide emission offsets during the last stages of emission reductions. We found that the emissions offsets provided by BECCS were much more valuable than the energy, suggesting that other sequestration options such as improved land management that increases soil carbon deposition could be a particularly valuable part of an economy-wide portfolio.

We started this research in the early days of the natural gas boom caused by widespread use of hydraulic fracturing. As data emerged on potentially high methane leakage rates in the natural gas supply chain, we investigated how leakage impacts roles Natural Gas (NG) can play in a low emission power grid. We found that leakage rates significantly reduce the use of NG as a direct substitute for coal, but have a smaller impact on the use of combustion turbines for reserves and peaking capacity. Higher leakage rates increase electricity costs in optimal solutions by an average of 1.3% ±0.068 and decrease NG consumption by 18% ±0.55 for each percentage point increase in the leakage rate in the next decade.

Increased leakage can increase or decrease the use of NG to complement renewables, depending on the emissions cap context and technological alternatives. In the 2020 and 2030 timeframes under moderate emission caps, higher leakage rates prompt the installation of more renewables and prompt NG Combined Cycle Gas Turbines (CCGT) to shift from baseload operation to running as-needed to complement renewables. In the 2030 timeframe, higher leakage often prompts installation of new NG Combustion turbines with Compressed Air Energy Storage, which is used to complement variability from renewable resources within a day. Scenarios that include low-cost battery storage or low-emission baseload options of Coal CCS or Nuclear have less Compressed Air Energy Storage installed in the 2030 timeframe because these technologies provide alternate emission reduction paths. In the 2040 and 2050 timeframes with tighter emission caps, NG is already used primarily to complement renewables and higher leakage rates tend to decrease its use in any role.

Throughout this process, I made significant advancements to Switch as an analytical tool for collaborative work by interdisciplinary research teams. I initially increased the usability and lowered the learning curve while training colleagues who lacked computer science backgrounds, as well as developing execution workflows to increase reproducibility and leverage high performance workstations and computing clusters. I played a crucial role in developing detailed databases to describe the WECC electricity grid and calculating renewable energy potential at a high geographic and temporal resolution over a large area. I developed new techniques for describing policies and tracking both the renewable fraction and emission intensity of electricity. I developed techniques for simulating grid dispatch of investment portfolios on ~100x as many timepoints to better estimate reliability, costs and emissions. I used that instrumentation ability to improve sampling methods and solution quality. Interviews with current and potential users indicated a need for a completely open source software stack, streamlined workflows for data ingestions and processing, as well as a graphical front-end to complement the command line interface. These usability enhancements are the subject of ongoing and future work.

Overall, this open collaborative approach has proven quite successful. We trained four other research teams on two campuses to develop versions of this model for China, Chile and Nicaragua and to conduct a detailed systems-level analysis of Carbon Capture and Sequestration technologies. Those efforts led to recognition by the United Nations during the 2014 Climate Summit. The complete text of this dissertation is freely available online through the University of California Berkeley though other organizations may distribute this text behind a paywall. We have developed partnerships with a second academic campus, a consulting firm and Google who are all contributing to a new implementation of Switch in a completely open source software stack that supports stochastic programming and decomposition (Pyomo). We hope that the new version can serve as a open platform for evaluating and comparing research methodologies as well as supporting investment planning and policy analysis for consulting firms, government agencies, academics, utilities and NGOs.