AcknowledgmentsThe editors of this report would like to acknowledge the contributions of a variety of participants, funders, and reviewers. Specifically, we would like to acknowledge the interest and support of Eric Smith of the U.S. Environmental Protection Agency (EPA) and key Department of Energy (DOE) Office of Energy Efficiency and Renewable Energy (EERE) staff, including Darrel Beschen, Sam Baldwin, and Michael Leifman (now with GE Energy). The editors would also like to thank Fred Beck of Sentech for his input.We would like to also thank several National Renewable Energy Laboratory (NREL) staff members for iterative and comprehensive reviews, including Doug Arent, Jeff Logan, Jim Newcomb, Gian Porro, and Michelle Kubik.Finally, we would like to thank our co-authors and other participants for their extensive contributions and engagement over a lengthy period of time; many participated without specific funding support. See the table of participants on Page 4 for a complete list. The study demonstrates that:
List of Acronyms• Different models and different technology and market assumptions can lead to widely different predictions of system outputs.• Even when technology and market assumptions are aligned as closely as possible, substantive differences still remain.To enable a comparison among various energy models, the group decided on a common scenario that all of the models could address. The group selected a penetration goal of 20% renewable energy generation in the electric sector by 2025, and conducted two broad sets of model runs:
2• A group of unaligned Base Case runs where modelers were allowed to use their own standard input assumptions including those for technology costs, fuels costs, and physical resources to achieve the target.• A group of aligned Tier 1 Case runs, where future technology and fuel costs, financial assumptions, and even resource supply curves were aligned to the extent possible to achieve the goal. This was done to separate the impact of inputs from structural differences in the models. This alignment will likely not happen in typical model use.We found that in both the aligned and unaligned cases, there was significant difference in the estimated output metrics, although the difference in predicted outcomes narrowed in the aligned case. Our analysis suggests that:• Due diligence needs to be exercised by policy-and decision-makers when presented with findings from a single model. Assumptions and model choices can lead to significantly different outcomes. For example, simple choices in future capital costs may result in a particular technology appearing dominant or marginal. Similarly, different models using identical technology and market assumptions might predict substantively different outcomes due to their structural differences. • Where possible, a variety of models using similar assumptions should be used to give the decision-maker a sense of differences in outcomes that reflect inherent uncertainties in the models, recognizing that some models are better suited to resolving...
