s1. further background on models for long-term …€¦ · web viewfor trade across the different...

13
Supplementary Online Material: Representing Power Sector Detail and Flexibility in a Multi-sector Model Marshall Wise 1 , Pralit Patel 1 , Zarrar Khan 1 , Son H. Kim 1 , Mohamad Hejazi 1 , Gokul Iyer 1 S1. Further background on models for long-term power sector planning Analytical tools and frameworks for long-term power sector planning vary along many dimensions. Here we discuss three dimensions: i.) spatial and temporal resolution; ii.) sectoral resolution and representation of interactions of power sector with other sectors of the economy; iii.) assumptions about foresight. For a detailed review of energy models see refs. [1-3]. A summary of the comparison of the improved version of GCAM-USA with other models in the literature across different dimensions is presented in Table S1. i. Spatial and temporal resolution: Models vary widely in terms of spatial and temporal resolutions. Multi-sector human-Earth systems models (also refereed to as integrated assessment models) resolve the power sector at the scale of countries or large world regions. In contrast, energy systems models of national scope include subnational detail (see Table S1). Likewise, the temporal scope of multi-sector human-Earth systems models is typically several decades into the future with annual resolution and limited representation of sub- annual processes and dynamics. In contrast, national energy systems models tend to include sub-annual detail. The level of spatial and temporal resolution used in these models could have implications for process detail. For example, too little spatial resolution can fail to capture important correlations or the full distribution of high- and low- quality variable renewable energy resources. Previous studies have found that aggregating renewable energy resources to coarser spatial resolution could lead to higher system costs because the highest-quality resource sites are averaged with mid- or low- quality sites. In contrast, aggregating transmission could lead to lower costs because lower-cost but more remote resources can contribute to the system with fewer transmission limitations [4]. Thus, resolution can have a substantial impact on model results. However, an important tradeoff for 1

Upload: others

Post on 08-Aug-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

Supplementary Online Material: Representing Power Sector Detail and Flexibility in a Multi-sector ModelMarshall Wise1, Pralit Patel 1, Zarrar Khan1, Son H. Kim1, Mohamad Hejazi1, Gokul Iyer1

S1. Further background on models for long-term power sector planning

Analytical tools and frameworks for long-term power sector planning vary along many dimensions. Here we discuss three dimensions: i.) spatial and temporal resolution; ii.) sectoral resolution and representation of interactions of power sector with other sectors of the economy; iii.) assumptions about foresight. For a detailed review of energy models see refs. [1-3]. A summary of the comparison of the improved version of GCAM-USA with other models in the literature across different dimensions is presented in Table S1.

i. Spatial and temporal resolution: Models vary widely in terms of spatial and temporal resolutions. Multi-sector human-Earth systems models (also refereed to as integrated assessment models) resolve the power sector at the scale of countries or large world regions. In contrast, energy systems models of national scope include subnational detail (see Table S1). Likewise, the temporal scope of multi-sector human-Earth systems models is typically several decades into the future with annual resolution and limited representation of sub-annual processes and dynamics. In contrast, national energy systems models tend to include sub-annual detail. The level of spatial and temporal resolution used in these models could have implications for process detail. For example, too little spatial resolution can fail to capture important correlations or the full distribution of high- and low- quality variable renewable energy resources. Previous studies have found that aggregating renewable energy resources to coarser spatial resolution could lead to higher system costs because the highest-quality resource sites are averaged with mid- or low-quality sites. In contrast, aggregating transmission could lead to lower costs because lower-cost but more remote resources can contribute to the system with fewer transmission limitations [4]. Thus, resolution can have a substantial impact on model results. However, an important tradeoff for improved resolution is computational and data requirement. Hence, most models do not represent investment and dispatch decisions across a full year. Instead, representative time intervals or “time slices” are typically used. Models use different methods to select individual hours or representative time steps from a full year of data including clustering techniques [5, 6]. An important shortcoming of using the time slice approach is the non-chronological nature of the resulting dispatch segments. This creates difficulties in modeling issues such as energy storage and demand-side management.

ii. Sectoral resolution: The evolution of the power sector, including investments and operation depends on influences from other human and natural systems including land-use, water, climate, and economy. For example, temperature extremes could result in high demands resulting in reserve capacity requirements [7]. Another example is the effect of water constraints on cooling water availability which could impact power plant operations [8]. Models differ in their ability to capture such influences. By definition, multi-sector human-Earth systems models capture interactions across sectors much more dynamically and endogenously compared to power sector only models. Ultimately, the degree of spatial, temporal, and sectoral resolution in a model depends on the research question and tradeoffs with computational effort [9].

iii. Assumptions about foresight: Models differ in their assumptions about foresight. Some models assume perfect foresight in their intertemporal optimization of management decisions (e.g. US-REGEN [10]) whereas other models use recursive-dynamic formulations or other

1

Page 2: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

approaches with more myopic decision-making (e.g. GCAM-USA [11, 12], and ReEDS [13]). In reality, decision makers do not have full information about future costs, and prices. This can often result in near-term decisions being weighted more since there could be more information available [14]. Models with myopic foresight typically assume that decisions made in a model period are dependent on conditions and prices in the current or prior periods. In contrast, models with perfect foresight assume that all future information is exact and simultaneously available for the whole time horizon to be modeled. Reality lies somewhere in between perfectly myopic and complete foresight. Modeling studies typically address this uncertainty in future decision-making by conducting multi-model exercises (see for example, Energy Modeling Forum studies [15, 16]) or exploring alternative scenarios.

S2. Representation of electricity trade in GCAM-USA

We do not represent sub-annual dynamics in electricity trade, but rather represent trade only at the annual temporal scale. As described earlier, we combine the states into 15 electric grid regions to reflect electricity market and planning areas (Figure S5). Within a grid region, we assume unconstrained trade within grid regions in all load segments and therefore common prices across these states. For trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy of nested non-linear equations (similar to equation 1 above)1. Net interregional trade is calibrated to historical levels to reflect existing economic conditions as well as implied physical transmission capability. In future modeling periods, trade can change from calibrated levels as relative regional electricity prices change. For example, a region that sees a relative price increase due to demand growth, cost increase, or policy change can import more from other regions. The nonlinear formulation means that the relative ease or elasticity of expanding these imports in a region tightens as the share of imports increases from the calibrated historical shares. That is, an increasing differential in regional prices is required to expand trade, reflecting an increasing marginal cost of building and maintaining expanded trade.

1 In contrast, linear programming models set fixed constraints on trade representing transmission capacity limits.

2

Page 3: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

S3 Additional Figures

Figure S1. Capacity additions (investments) in the U.S. power sector by technology in the various scenarios explored in this paper.

(a) (b)

Figure S2. U.S. a.) Electricity generation and b.) Capacity by fuel in the nine dispatch segments in 2050 in the Reference scenario.

3

Page 4: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

Generation

Capacity

Figure S3. Electricity generation and capacity in the 25 dispatch segments in 2030 in the fifteen grid regions in the Reference scenario.

4

Page 5: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

Figure S4. Dispatch curves for Texas and Central East grids in 2050 in the Reference scenario.

5

Page 6: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

Figure S5. Mapping of states in GCAM-USA to grid regions

Figure S6. Final energy by fuel in the U.S. in the various scenarios explored in this paper.

6

Page 7: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

Figure S7. Electricity consumption in 2030 in the various scenarios explored in this paper.

7

Page 8: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

Figure S8. Changes in U.S. electricity generation (panels a, b, c) and capacity (panels d, e, f) under the different natural gas price scenarios considered in this study relative to the Reference scenario using the version of the GCAM-USA model without the separate dispatch and investment ability discussed in the paper. The version of the model used for this exercise corresponds to version 4.4.1.

8

Page 9: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

S4 Comparison to Other Models

Table S1 Comparison of improved version of GCAM-USA with other models in the literature

Model [Reference]

Spatial scope

Spatial resolution

Temporal scope

Temporal resolution in power sector

Foresight

Sectoral resolution

ReMIND [17]

Global 11 regions (5 induvial countries and 6 aggregated regions)

2005-2100

Load duration curve divided into four bands

Perfect Energy supply, transformation, and demand, land-use; economy

IMAGE [17]

Global 26 regions 1970-2100 in 5-year time step

Load duration curve divided into 20 subdivisions

Myopic Energy supply, transformation, and demand, land-use, and agriculture; economy

NEMS [18] U.S. Generation of electricity is accounted for in 22 supply regions in the U.S.

2010-2050 in annual time-step

Load duration curve divided into 9 segments

Perfect Energy supply, transformation, and demand; economy

ReEDS [19]

Contiguous U.S., and some representation of Canada, Mexico

134 load balancing areas covering contiguous U.S.

2010-2050 in 2-year increments

17 time slices Myopic. Perfect foresight for natural gas and CO2 prices

Power sector only (supply)

US-REGEN [20]

Contiguous U.S.

Customizable regions based on state boundaries; default 15

Three-year increments through 2030 and five-year

Customizable; typically 100+ “representative hours”

Perfect Energy supply and demand

9

Page 10: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

increments through 2050

GCAM-USA [Version used in the current study]

Global and U.S.

50 states and D.C. in the U.S. along with 31 regions outside of the U.S.

2010-2100 in 5-year or 1-year time step

Load duration curve with 25 dispatch segments and 4 investment segments

Myopic Energy supply, transformation, and demand; agriculture, and land-use; water; climate; economy.

References

1. Ringkjøb, H.-K., P.M. Haugan, and I.M. Solbrekke, A review of modelling tools for energy and electricity systems with large shares of variable renewables. Renewable and Sustainable Energy Reviews, 2018. 96: p. 440-459.

2. Collins, S., et al., Integrating short term variations of the power system into integrated energy system models: A methodological review. Renewable and Sustainable Energy Reviews, 2017. 76: p. 839-856.

3. Pfenninger, S., A. Hawkes, and J. Keirstead, Energy systems modeling for twenty-first century energy challenges. Renewable and Sustainable Energy Reviews, 2014. 33: p. 74-86.

4. Cole, W., et al., Variable renewable energy in long-term planning models: a multi-model perspective No. NREL/TP-6A20-70528). National Energy Technology Lab.(NETL), Pittsburgh, PA, and Morgantown, WV (United States). 2017.

5. Blanford, G.J., et al., Simulating Annual Variation in Load, Wind, and Solar by Representative Hour Selection. The Energy Journal, 2018. 39(3).

6. Getman, D., et al., Methodology for Clustering High-Resolution Spatiotemporal Solar Resource Data. NREL/TP-6A20-63148. Golden, CO: National Renewable Energy Laboratory. https://www.nrel.gov/docs/fy15osti/63148.pdf. 2015.

7. Auffhammer, M., P. Baylis, and C.H. Hausman, Climate change is projected to have severe impacts on the frequency and intensity of peak electricity demand across the United States. Proceedings of the National Academy of Sciences, 2017. 114(8): p. 1886–1891.

8. Liu, L., et al., Implications of water constraints on electricity capacity expansion in the United States. Nature Sustainability, 2019. 2(3): p. 206-213.

9. Calvin, K. and B. Bond-Lamberty, Integrated human-earth system modeling—state of the science and future directions. Environmental Research Letters, 2018. 13(6): p. 063006.

10. Electric Power Research Institute, US-REGEN Model Documentation. https://www.epri.com/#/pages/product/000000003002010956/. 2017.

10

Page 11: S1. Further background on models for long-term …€¦ · Web viewFor trade across the different grid regions, we place the electricity markets of the grid regions in a hierarchy

11. Iyer, G., et al., GCAM-USA Analysis of US Electric Power Sector Transitions. http://www.pnnl.gov/main/publications/external/technical_reports/PNNL-26174.pdf. 2017, Pacific Northwest National Laboratory.

12. Iyer, G., et al., Measuring Progress from Nationally Determined Contributions to Mid-Century Strategies. Nature Climate Change, 2017. 7: p. 871–874.

13. Cole, W., et al., 2017 Standard Scenarios Report: A U.S. Electricity Sector Outlook. https://www.nrel.gov/docs/fy18osti/68548.pdf, 2017.

14. Keppo, I. and M. Strubegger, Short term decisions for long term problems – The effect of foresight on model based energy systems analysis. Energy, 2010. 35(5): p. 2033-2042.

15. Creason, J.R., et al., Effects of technology assumptions on US power sector capacity, generation and emissions projections: Results from the EMF 32 Model Intercomparison Project. Energy Econ, 2018. 73: p. 290-306.

16. Murray, B.C., et al., The EMF 32 study on technology and climate policy strategies for greenhouse gas reductions in the U.S. electric power sector: An overview. Energy Economics, 2018.

17. ADVANCE wiki, The Common Integrated Assessment Model (CIAM) documentation. http://themasites.pbl.nl/models/advance/index.php/ADVANCE_wiki. 2015.

18. U.S. Energy Information Administration, NEMS Documentation. https://www.eia.gov/outlooks/aeo/nems/documentation/ 2018.

19. Cohen, S., et al., Regional Energy Deployment System (ReEDS) Model Documentation: Version 2018. Golden, CO: National Renewable Energy Laboratory. NREL/TP-6A20-72023. https://www.nrel.gov/docs/fy19osti/72023.pdf. 2019.

20. EPRI, US-REGEN Model Documentation, https://www.epri.com/#/pages/product/3002010956/?lang=en-US. 2018.

11