The economics of large-scale wind power in a carbon constrained world
Introduction
In 2002, the wind power industry generated total sales of $5.8 billion, with over 32 GW of wind capacity installed worldwide (BTM Consult, 2003). At good sites, the average cost of wind is currently 4– without credits or subsidies, and advances in turbine design may plausibly reduce the cost to in the next two decades. Although wind energy currently represents about 0.1% of total global electricity (Sims et al., 2003), it has the fastest relative growth rate of any electric generating technology: capacity has increased by roughly 32% annually for the 5 years ending in 2002 (AWEA, 2003). The absolute annual growth of wind power generation now exceeds that of hydro, but is still an order of magnitude smaller than for natural gas fired electricity. Several analyses suggest that wind could feasibly serve at least 10–20% of electricity demand globally (EWEA, 2003) or regionally (Ilex and Strbac, 2002, Gardner et al., 2003) within a few decades. The rapid growth of wind capacity and the aggressive projections of future growth are driven by two factors: the declining cost of wind technology and strong policy incentives for wind development.
Two factors—the spatial distribution and intermittency of wind resources—raise the cost of large-scale wind above the average cost of electricity from a single turbine. Additional costs arise from long distance electricity transmission (to compensate for mismatch between the spatial distribution of wind resources and demand) and backup capacity and/or storage systems (to compensate for the mismatch in temporal distribution of supply and demand). While these costs arise at any scale, their influence on the economics of wind-power grow rapidly as wind serves a larger fraction of demand.
We investigate the economics of using wind to reduce emissions in future electric power systems. Our focus is on the utilization of wind to serve more than a third of electricity demand by 2030 in a regulatory environment shaped by carbon constraints. Given the uncertainty about the regulatory and technological paths that the electricity industry might take over the next quarter century, we do not make predictions about the likely mix of generating technologies in 30 years, nor the temporal evolution of the generating mix. We aim to understand the cost-effectiveness of using wind to mitigate carbon emissions in a carbon constrained world, while accounting for the remote location and intermittency of wind resources. We employ a greenfield optimization model that rests on a time resolved simulation of wind power, demand, and storage in order to determine the optimal wind, gas turbine, storage, and transmission capacities in a hypothetical system under a carbon tax.
The timing of serious regulatory constraints on emissions remains profoundly uncertain. When such constraints arrive, the electric sector will likely need to deliver deeper proportional reductions in emissions than elsewhere in the economy. There are several reasons to expect that the electricity sector will be a key target for carbon mitigation. Centralized ownership and management of electric power plants, which are the largest and most manageable point sources of emissions, make regulation easier to implement in an industry that already has considerable experience with the regulation of emissions (Johnson and Keith, 2004). If serious efforts are made to slow climate change, then the US electric sector will likely need to cut emissions in half within the next quarter century. Wind power may play a pivotal role in reducing emissions from electric power generation.
There is no panacea for eliminating emissions in the electricity sector. Because wind is a viable emissions-free technology, a more accurate assessment of the cost of mitigating electric sector emissions using wind is important to the economics of climate change mitigation. Other options include fuel switching to less carbon-intensive fuels, improved efficiency (both demand- and supply-side), carbon capture and sequestration (CCS), biomass, nuclear, and photovoltaics. Each of these alternatives possesses a unique set of benefits, limitations, and costs. Although our analysis focuses on wind, we recognize the potential efficacy of these other options.
The rapid worldwide growth in wind capacity has been driven by environmentally motivated taxes, credits, and other regulatory incentives. Absent such incentives, we do not expect that wind will achieve substantial penetration into worldwide electricity markets, despite the continued declining costs of wind turbines, in part because of the costs imposed by remoteness and intermittency at high penetration levels. We assume that the most important driver for future wind development will be a constraint on carbon emissions. Under a strong carbon constraint, it is likely that wind will compete effectively with other means of reducing electric sector carbon emissions such as coal with CCS or nuclear. Despite assertions to the contrary (NREL, 2002, UCS, 2003), wind is unlikely to become a competitive means to achieve reductions in air pollution or to enhance energy security. If air pollution reduction is the goal, then deep reductions in air pollutants can be achieved by retrofits to existing coal facilities at costs of order (Rubin et al., 1997). If energy security is the driving concern, then for many nations, coal provides sufficient security. The reserve/production ratio for coal is about 200 years globally, and 250 years in the US (BP, 2003).
Understanding the long-term role of wind in a constrained world requires us to bridge two domains of analysis. First, there is a rich set of analyses that examine the integration of wind power into existing electricity transmission systems and their associated electric markets. Such analyses generally look no more than two decades ahead and/or assume that much of the existing electric power infrastructure remains in place (e.g., Grubb, 1988, Hirst, 2001, Ilex and Strbac, 2002). Second, there is a similarly rich set of analyses that examine the long-term economics of the -climate problem. These include energy models of the kind that participate in the Energy Modeling Forum, and Integrated Assessment models that embed energy system models with models of the climate system and the impacts of climate change to assess climate policy. These models often examine a century long time horizon, and include representations of technological change and economic growth. While these models often include wind, they cannot readily capture the dynamics of load and dispatch in electric power systems and markets (e.g., Edmonds et al., forthcoming). The aim of our analysis is to bridge the gap between these intellectual domains by simulating large-scale wind in a greenfield electric power system. In addition, our analysis provides cost estimates (in the form of supply curves) of mitigating carbon emissions with wind at high penetration levels that could be used in developing more accurate treatments of wind in long-duration comprehensive models aimed at understanding the cost of mitigating emissions.
Analysis by Cavallo (1995) addresses the issue by estimating the cost of “baseload” wind (a wind-storage system with 90% capacity factor) at . Because Cavallo's analysis focuses on a specific case study of a Kansas windfarm connected to southern California via a 2000 km HVDC line, it is difficult to extrapolate the results to scenarios that include multiple wind sites, where the utilization of weakly correlated wind sites might improve the economics.
Recent analysis by Ilex Energy Consulting (2002), examined the balance of system costs incurred by renewables serving 20% and 30% of electricity demand in Great Britain. In the North Wind scenario with high demand, the additional system cost due to wind energy serving 30% of electricity demand is (Ilex and Strbac, 2002). However, the analysis does not include the cost of the wind turbines or the cost of new transmission to tie the wind farms to the grid—only the system costs incurred for grid reinforcement, managing transmission losses, balancing, and security.
The National Renewable Energy Laboratory is developing a model called the Wind Deployment Systems Model (WinDS), a multi-regional, multi-time-period, GIS and linear programming model. Preliminary results indicate that in the base case (with infinite extension of existing regulatory incentives) wind capacity may reach several hundred GW in the next 50 years (Short et al., 2003).
In contrast to the studies mentioned above, we focus on the cost-effectiveness of large-scale wind in meeting a constraint. In order to do this, we simulate the interaction of several large wind farms and a time-varying demand in a greenfield scenario, where wind, storage, transmission lines, and gas turbines are optimized to meet load on an hourly basis. Our analysis builds on Cavallo's work by including multiple wind sites in order to quantify the benefit of geographically dispersed wind farms, which exhibit greater aggregate reliability by exploiting less correlated wind patterns. At the same time, our analysis is meant to be transparent and generalizable, in contrast to the Ilex analysis (2002) and NREL's WinDS model which are detailed policy analyses with a strong national focus.
The following section describes the challenges posed by the spatial distribution and intermittency of wind resources. Section 3 describes the structure, assumptions, and results of our model. Section 4 provides a more detailed analysis of the role of storage in large-scale wind systems under a carbon constraint. Finally, Section 5 provides a summary of the model results and draws conclusions for the future of large-scale wind energy in the long-term.
Section snippets
Intermittency and location
The intermittency of wind energy can affect an electricity grid on timescales ranging from less than a second to days. Three timescales concern system operators on a day-to-day basis: minute-to-minute, intrahour, and hour- to day-ahead scheduling. System operators employ an automatic generation control (AGC) system to manage minute-to-minute load imbalances—an ancillary service known as regulation. Operating reserve, which consists of spinning and non-spinning reserves, represents capacity that
Model
The purpose of this optimization model is twofold: (i) to provide an economic characterization of large-scale wind when intermittency and remoteness cannot be ignored, and (ii) to determine the cost of carbon mitigation of wind at different levels of penetration by constructing supply curves. The greenfield model minimizes the averaged delivered electricity cost by adjusting geographically dispersed wind turbine arrays, a storage system, and backup gas turbines to meet a time-varying load under
Exploring the benefits of CAES
The absence of CAES capacity in Fig. 2B and the utilization of CAES only at high carbon taxes in Fig. 2A is an intuitively surprising result. Residual emissions generated by the CAES system handicap its performance under a carbon tax, such that CAES does not compete effectively with GT and GTCC capacity. To scrutinize CAES performance under a variety of assumptions, a reduced form optimization model was constructed. Rather than embedding a simulation of wind power within the optimization, the
Conclusions
The model presented here allows us to estimate the cost of using large-scale wind to achieve deep cuts in emissions by optimizing distributed wind sites, transmission lines, storage, and gas turbines to mitigate the problems posed by remoteness and intermittency. While the model is idealized, we can nevertheless draw several interesting conclusions about the use of large-scale wind.
First, assuming comparatively low costs for wind turbines and a low discount rate of 10%, the average cost of
Acknowledgements
The authors wish to thank Alex Farrell of UC Berkeley, David Denkenberger and Jeffrey Greenblatt of Princeton, as well as Granger Morgan, Jay Apt, Lester Lave, and Hisham Zerriffi of Carnegie Mellon University for their insights. This research was made possible through support from the Carnegie Mellon Electricity Industry Center (CEIC). CEIC is supported by the Alfred P. Sloan Foundation and the Electric Power Research Institute.
References (40)
- et al.
Fossil electricity and sequestrationhow natural gas prices, initial conditions and retrofits determine the cost of controlling emissions
Energy Policy
(2004) Reliability of distributed wind generators
Electric Power Systems Research
(1979)- et al.
Carbon emission and mitigation cost comparisons between fossil fuel, nuclear and renewable energy resources for electricity generation
Energy Policy
(2003) - et al.
Spatial and temporal distributions of U.S. winds and wind power at 80 m derived from measurements
Journal of Geophysical Research
(2003) - et al.
Abatement by the combustion of -rich fuels in gas turbines greenhouse gas control technologies
- AWEA, 2003. Global wind energy market report. American Wind Energy Association, Washington, DC, accessed 5 February...
- Baxter, R., Makansi, J., 2003. Energy storage: an innovative answer to a few of today's challenges. Energy Pulse,...
- Borroughs, S., Bauer, S., 2001. Solution to some of Country's Energy Woes might be little more than Hot Air. Sandia...
- BP, 2003. BP Statistical Review of World Energy. British Petroleum, Egham,...
- BTM Consult, 2003. International Wind Energy Development: World Market Update 2002. BTM Consult, Ringkøbing,...
High-capacity factor wind energy systems
Journal of Solar Energy Engineering
Study of Electric Transmission in Conjunction With Energy Storage Technology
Steady as she blows
IEEE Spectrum
The Sound of Power Politics
Cited by (193)
Towards a 3D-resolved model of Si/Graphite composite electrodes from manufacturing simulations
2021, Journal of Power SourcesEffects of wind power intermittency on generation and emissions
2019, Electricity JournalScale evolution, intermittency and fluctuation relations in the near-wake of a wind turbine array
2019, Chaos, Solitons and Fractals