TODAY’S STUDY: RENEWABLES AND GRID RELIABILITY
Meeting Load with a Resource Mix Beyond Business as Usual; A Regional Examination of the Hourly System Operations and Reliability Implications for the United States Electric Power System with Coal Phased Out and High Penetrations of Efficiency and Renewable Generating Resources
Tommy Vitolo, Geoff Keith, Bruce Biewald, Tyler Comings, Ezra Hausman, and Patrick Knight, April 17, 2013 (Synapse Energy Economics)
A “business as usual” strategy for the U.S. electric power industry, wherein the country continues to rely heavily on coal and other fossil fuels to meet its energy needs, is not tenable if we are to achieve substantial reductions in greenhouse gas emissions over the next several decades. In 2011, Synapse prepared a study for the Civil Society Institute,
Toward a Sustainable Future for the U.S. Power Sector: Beyond Business as Usual 2011 (BBAU 2011), that introduced a “Transition Scenario” in which the United States retires all of its coal plants and a quarter of its nuclear plants by 2050, moving instead toward a power system based on energy efficiency and renewable energy. Synapse’s study showed that this transition scenario, in addition to achieving significant reductions in emissions of CO2 and other pollutants, ultimately costs society less than a “business as usual” strategy—even without considering the cost of carbon. BBAU 2011 projected that, over 40 years, the Transition Scenario would result in savings of $83 billion (present value) compared to the business as usual strategy.
As part of this lower-cost and low-emissions strategy, the Transition Scenario included large amounts of renewable energy resources with “variable output,” such as wind and solar. Without the inclusion of these resources, it will be difficult or impossible to reduce electric-sector greenhouse gas emissions to the levels necessary to materially mitigate our contribution to dangerous climate change.
While the need for variable-output resources is well defined, questions have been raised about the impact of large-scale wind and solar integration on electric system reliability.1 In light of this important concern, Synapse paid careful attention to the amount of wind and solar in each region when designing the Transition Scenario for BBAU 2011, taking steps to ensure that the projected regional resource mixes could respond to all load conditions. These steps included: • improving the capability of the transmission system to handle large interregional power transfers; • ensuring that regions with high levels of variable generation also had high levels of flexible generation and capacity; • adding storage capacity in regions with high levels of wind generation;2 • strengthening the capability and flexibility of electric systems through transmission and distribution investments; and • developing robust demand-side management resources.
Our current study takes the analysis deeper, in order to explore the extent to which the Transition Scenario’s resource mixes for 2030 and 2050 are capable of meeting projected load for each of the ten studied regions—not just during peak demand conditions, but in every hour of every season of the year as consumers require. Using a simplified hourly dispatch model along with empirical load and resource output profiles, we assess the ability of the projected mix in each region to meet load under the varying conditions throughout a day, season, and year. An important limitation of the dispatch model is that it does not include the interregional transfers that were a fundamental part of the resource mix under BBAU 2011, as these have not been defined on an hourly basis. These transfers are an important part of the Transition Scenario for both economic and reliability reasons, and indeed we find that under certain extreme conditions, it is impossible to balance each region in isolation. Nonetheless, our analysis shows that the regional Transition Scenario resource mixes would be capable of meeting load for almost all hours of the year in each region, and that a combination of interregional transfers, local storage, and demand response would be more than adequate to provide a high level of reliability.
This analysis, along with BBAU, is solely based on today’s existing technology. We do not expect that the optimal sustainable electricity future for the United States will look exactly like our Transition Scenario, as we anticipate that changes in the technology and economics of carbon- free generation and energy storage will produce options that today would seem unachievable. What we demonstrate in this report is that strategies to address one of the most pressing challenges faced by our species and our planet are already not only achievable, but cost effective. Future developments will only improve this potential—it is up to policymakers to make this potential a reality…
Summary of Findings
With few exceptions, this study finds that BBAU 2011’s Transition Scenario resource mixes, based entirely on existing technology and operational practices, are capable of balancing projected load in 2030 and 2050 for each region—in nearly every hour of every season of the year. Of course, any viable scenario must be based on much higher levels of reliability, such as a one-outage-in- ten-years standard currently used throughout the United States today. Thus we focus here on any hours with an energy imbalance, either as “unusable surplus” or shortages, to investigate their implications for the feasibility and implementation requirements of the Transition Scenario. This analysis highlights the ways in which interregional cooperation, followed by improvements in technology such as energy storage systems, can provide very high levels of reliability under the Transition Scenario..
In some cases, additional research and/or modifications to the resource mixes posited by BBAU 2011 may be warranted. Discussed in Section 3 of this report, these cases may include the energy shortfalls observed in the southeast and western regions in the summer and winter seasons, and the energy surpluses that occur in Texas and other regions in the shoulder seasons.
As noted above, the BBAU resource mix generally should be seen as an illustrative example, and was never identified as an “optimal” scenario. Integration analysis far beyond that presented here will be an integral part of defining the best combination of resources to provide reliable electric service in a carbon-constrained world. The earlier this sort of in-depth analysis is undertaken, the more options will be available for meeting resource adequacy requirements in a cost-effective way.
This study suggests that it will be feasible to reliably integrate the high levels of zero-carbon energy called for by the Transition Scenario, whether or not this scenario will ultimately provide the most cost effective or elegant nationwide low-carbon energy solution. Achieving this level of integration will likely require incremental improvements in technology and operational practices, including continuation of the current trend toward better interregional coordination. In contrast, the alternative—continuing to rely on increasing combustion of fossil fuels and to bear the growing toll on natural resources and the Earth’s climatepresents far more daunting technical, economic, and social challenges to human and environmental welfare.
The “Beyond Business as Usual” Study
This study relies heavily on a November 2011 Synapse study for the Civil Society Institute titled Toward a Sustainable Future for the U.S. Power Sector: Beyond Business as Usual 2011 (BBAU 2011). BBAU 2011 evaluated and compared two scenarios:
1) Business as Usual (BAU): Under this scenario, which was based on the U.S. Energy Information Administration’s Annual Energy Outlook (AEO) 2011 modeling work, the country continues to rely on fossil and nuclear generation to meet its energy needs, and electric-sector carbon dioxide (CO2) emissions continue to increase.
2) Transition Scenario: Under this scenario, the country moves toward a power system based on efficiency and renewable energy, and CO2 emissions are reduced substantially. In the Transition Scenario, all U.S. coal-fired power plants are retired, along with nearly a quarter of the nation’s nuclear fleet, by 2050.
For BBAU 2011, Synapse estimated the net costs and benefits of the Transition Scenario relative to BAU using a spreadsheet model that accounted for generating capacity, energy, fuel use, costs, emissions, and water use. Synapse performed the analysis on a regional basis, with the country divided into ten regions aggregated from the 22 regions used in AEO 2011 as shown in Figure 3. For each region, Synapse ensured that there was sufficient generating capacity in both the BAU and Transition scenarios, and that there was a generally reasonable mix of energy sources in each region from the perspective of power system operation.
The analysis for the current study is focused on these same ten regions.
For most of our technology cost and performance assumptions, we relied on the AEO 2011 data (U.S. EIA 2011). If judged to be more accurate than AEO 2011, other data sources were used for some technologies.
BBAU 2011 found that the Transition Scenario was significantly less expensive than the BAU Scenario—saving a present value of $83 billion over 40 years. This finding was particularly striking, given that the BAU Scenario included no carbon costs or carbon reductions. If the cost of carbon reductions (or the societal cost of continued emissions) were included in the BAU Scenario, the savings provided by the Transition Scenario would have been far higher.
Synapse included a large amount of zero carbon, variable output resources—i.e., wind and solar—in the Transition Scenario. In designing this scenario, Synapse paid careful attention to the wind and solar energy potential in each region, and attempted to ensure that the projected resource mixes and interregional transfers were likely to be capable of meeting all load conditions. These steps included: ensuring that regions with high levels of variable generation also had high levels of flexible generation and capacity; adding storage capacity in regions with high levels of wind generation; strengthening the capability and flexibility of electric systems through transmission investments; and including the cost of implementing robust demand response programs. We also noted that trends in system operation—such as consolidation of balancing areas, and increased information sharing—were likely to facilitate the integration of variable resources under either scenario.
Our present study takes the analysis deeper to explore the extent to which the Transition Scenario resource mixes for 2030 and 2050 meet projected hourly load for each of the ten regions.
Integrating Variable-Output Generating Resources
Historically, grid operators have responded to real-time changes in demand by virtually instantaneous control of generating resources to maintain frequency and voltage, and to balance electricity supply and demand. Outside of scheduled maintenance outages and unforeseen events, such as the failure of a generating plant or the loss of a transmission line, operators have assumed that generators are available and reliable, and that demand is fairly predictable— especially if weather conditions are known. Integrating high levels of variable-output resources into the electric grid will require a significant shift in perspective from grid operators (DOE 2008).
While variable-output generators cannot be directly controlled by the operator, they provide significant benefits including increased price stability and contributions to meeting peak demand (APS 2010). By reducing the usage of fossil fuels to produce electricity, solar and wind resources also provide significant benefits in terms of reducing an electric system’s greenhouse gasses, air pollutants, water usage, and solid waste.
The output of solar resources is dependent on the angle of the sun and the presence of clouds. Based on current scientific knowledge, we are able to forecast the angle of the sun with complete accuracy for centuries into the future. Using satellites and other meteorological tools, we can forecast the presence of clouds at a given location for several hours into the future.
Solar thermal resources—which use mirrors to focus sunlight to heat steam for a turbine—cannot operate without direct sunlight; however, they are often able to store heat and thereby continue generating electricity for several hours after dusk (DOE 2008). Solar photovoltaic (PV) resources, on the other hand, do not require direct sunlight to generate electricity, but offer no storage ability. They can be mounted in a fixed position, or can change their angle throughout the day to be optimally positioned with respect to the angle of the sun. Intermittent clouds introduce unpredictability for PV facilities, since they produce energy at lower levels if direct sunlight is not available. PV resources do not produce energy after dusk. Despite these constraints, both types of solar resources are beneficial to the electrical system, since optimal operating conditions with direct sunlight often coincide with summer peak demand (MIT 2012).
The output of wind resources is often characterized as being very unpredictable.
However, while the output of an individual wind turbine at any point in time is extremely difficult to predict, the output of a group of turbines becomes more predictable as the number of turbines and their geographic diversity increase. Individual turbines are sensitive to changes in wind strength which can be localized and short-lived, or broad-scaled and persistent. In contrast, large groups of turbines, such as wind farms, are less subject to local and short-lived variations, and regions with geographically diverse wind resources are even more robust (DOE 2011). Wind resources can also exhibit predictable seasonal and diurnal variations; turbines are typically more likely to run in the early morning and in the winter. Even if the wind is not coincident with peak demand, large- scale patterns provide predictability for balancing purposes (MIT 2012). Additionally, new wind forecasting tools are being developed to help system operators prepare for changes in wind production. The Electric Reliability Council of Texas (ERCOT), working with AWS Truepower and other third parties, has implemented a tool to provide useful 6-hour, 4-hour, and 2-hour-ahead wind power forecasts.
Today, unpredictable variations in load and in the output of variable-output resources are accommodated through the use of high flexibility resources including storage hydropower and flexible mid-merit and peaking gas units. The Transition Scenario was designed to include sufficient quantities of these resources to meet additional variable output generation. These resource types include: • Storage Hydro - Hydro facilities with reservoirs can be quickly ramped up or down in response to load, which is useful for complementing variable renewable generation (Denholm 2010). Today many of these facilities use their storage capability to generate as much electricity as physically possible during high-load and high-cost daytime hours and little or none overnight. • Combustion Turbine (CT) Peaking Units - Gas-fired combustion turbines can be ramped up or down quickly; however, they are also the least efficient and typically the most expensive generators to run (MIT 2012). Peaking units typically have a very short lead time for construction, and can be installed quickly to help meet expected growth in load or in the need for flexible generation. • Combined-Cycle Combustion Turbine (CCCT) Gas Plants - Gas-fired combined cycle plants provide a valuable mix of high efficiency and operational flexibility to complement variable resources (MIT 2012). They can ramp up and down quickly, and are more efficient and cost-effective to run than CT (peaking) units, as they require less natural gas per MW of output. However, they are more expensive to build and require longer construction lead-time than CTs.
Energy Storage exists today in the forms of pumped hydropower, compressed air storage, flywheels, and batteries. Thermal energy storage in buildings and industrial settings is also used today. Storage provides the ability to both absorb electricity during hours of surplus and to dispatch it as a generator at a later time. Energy storage will always involve some level of losses—for example, it takes more energy to fill a pumped hydro storage reservoir than can be recovered by releasing the water. Today’s advanced storage technologies, such as batteries and flywheels, are relatively expensive and limited in scale, and have thus been applied mostly for specialty applications. However, lower-cost energy storage is an area of very active research and development, including efforts to improve batteries, develop hydrogen production and storage, and implement end-use storage such as thermal storage in buildings, electric water heaters that can respond to system operator controls, and plug-in electric vehicles. Energy storage will almost certainly play an important role in any energy future with higher levels of renewable resources, because storage effectively converts intermittent energy generation to highly flexible dispatchable generation. This study assumed that future storage would have the same cost and efficiency structure as current storage; however, technological advancements will only improve the cost and performance of electrical storage over time.
Technologies Facilitating Integration
All of the aforementioned constraints and operating characteristics must be taken into account when integrating generating resources into the grid, in order to maintain the balance of generation of load.
High levels of variable-output generation (wind and solar) add another layer of complexity to the existing challengs of balancing generation and load in real time while ensuring high levels of reliability. Fortunately, a number of tools are available or under development to help grid operators more easily capture the benefits of variable generation while maintaining a reliable electric system. These tools include electricity and thermal energy storage (described above), extended use of demand response resources, and smart grid applications that can be used for load and frequency balancing (APS 2010; Denholm 2010; MIT 2012). The wider use of electric vehicles will also provide an opportunity for storage and load control to the grid (MIT 2012). As discussed above, geographic diversity and diversity of resource types over larger regions will naturally smooth out some of the variability and unpredictability associated with variable-output resources.
Finally, improved approaches to using existing, flexible resources such as storage hydro and gas, combined with better forecasting for variable resource output and real-time control, will substantially enhance the ability to accommodate high levels of variable-output renewable energy (Lew et al. 2010).
Dispatch Model Analysis
The purpose of this study was to determine whether and in how many hours of the year the BBAU 2011 Transition Scenario resource mixes for 2030 and 2050 resulted in insufficient electricity available to serve load, or an unusable surplus of electricity.
To perform this analysis, Synapse built a simplified hourly dispatch model based on hourly, regional matching of resources to load. Inter-hourly constraints such as generator ramping limitations are not considered, nor are local transmission constraints. The model does not explicitly model imports and exports between regions. Finally, the model does not consider the need for reserves or any other ancillary service. While a more comprehensive, in-depth dispatch modeling study might accommodate these dynamics and constraints better, we believe the analytical benefits would be illusory; they would be based on limitations and operational practices of today that are not likely to be characteristic of the future study years. On the other hand, the model does not model demand response as a resource. When dispatched, demand response resources allow the systems operator to shift the load curve in order to mitigate or eliminate energy imbalances. This, along with the exclusion of interregional transfers of power, renders the model relatively conservative for this analysis.
Hourly load data for each region was based on 2010 actual demand, and was adjusted— considering changes in demographics, wealth, and energy efficiency—so that the peak load and annual energy requirements closely matched those in the BBAU 2011 Transition Scenario. Data for these tasks were obtained from FERC 2011, NERC 2012, and U.S. EPA 2011. The generators used in the model came from the BBAU 2011 Transition Scenario. To model the hourly generation of variable resources, a number of National Renewable Energy Laboratory (NREL) studies and data sets were used (NREL PVWatts 2012, GE Energy 2010, and EnerNex Corporation 2011).
Order of Dispatch
For purposes of dispatching units in order of economic merit to meet load, generating resources were divided into four major dispatch categories: low-flexibility dispatchable generation (such as baseload nuclear and coal), variable resources, high-flexibility dispatchable generation, and storage. The model simulates unit commitment5 by looking ahead to the upcoming week (168 hours) to determine if coal or biomass resources would be needed to meet demand, or if they would be called on in the ordered dispatch frequently enough to justify being made available for the week. The model then calculates hourly load net of variable resource output to determine how much energy from conventional resources is required, if any. If variable resource output is too high relative to load, the model attempts to absorb the excess energy into available storage. If more energy is required, the model tries to meet load using the following resource ordering, using all available capacity in one before moving on to the next: storage hydro; coal (if available); biomass (if available); energy stored from any surplus in previous hours; CCCT gas, and then peaking gas. If all of those resources, when dispatched, still fail to meet load, any available emergency storage is called upon. The model assumes that, in an actual scenario like this, system operators would have anticipated the need for energy reserves,6 and would have prepared by storing surplus energy in the prior time periods.7 If the emergency storage is not sufficient to meet load, then a shortfall occurs.
Under realistic operating conditions, it is likely that techniques to shift load such as time-of-use pricing and thermal and chemical storage demand response would be employed, thereby reducing the extent of surpluses and shortages. Any shortfall would be met by importing energy from a neighboring region (as is commonly done for economic and reliability reasons today) or by the use of additional demand response. These resources are not available to the dispatch logic in our model, rendering the dispatch analysis conservative relative to the actual challenged likely to be faced by system operators…
Conclusions & Recommendations
With few exceptions, this study finds that BBAU 2011’s Transition Scenario resource mixes, based entirely on existing technology and operational practices, are capable of balancing projected load in 2030 and 2050 for each region—in nearly every hour of every season of the year. Of course, any viable scenario must undergo an extensive suite of analysis, including probabilistic electric system reliability modeling. This study highlights the ways in which interregional cooperation, along with improvements in technology such as energy storage systems, can provide very high levels of reliability under the Transition Scenario.
The primary limitation of this analysis is the lack of important resource options for balancing load— interregional transfers and demand response—that would almost certainly play a key role in a clean-energy future; and indeed that are in widespread use today, and that were an important element of the BBAU 2011 Transition Scenario. Use of these resources would almost certainly substantially reduce or eliminate regional imbalances, and would make system operations more efficient and economical. On the other hand, the fact that the regions were almost always able to balance load without these resources adds to our confidence in the capability of the Transition Scenario.
The BBAU 2011 Transition Scenario resource mix was never intended to be an “optimal” scenario. Our expectation is that improvements in technology and operational practices over the coming decades will eclipse the resource options and practices that we can envision today. Nonetheless, we believe that providing a comprehensive, feasible vision for a clean energy future (and highlighting the technological challenges such a future presents) is an important contribution to facilitating this crucial transition. The sooner that we undertake in-depth analyses of resource and integration needs, the more options will be available for meeting future resource adequacy requirements in a cost-effective way.
Although it is unlikely that BBAU 2011’s Transition Scenario will ultimately provide the most cost effective or elegant nationwide low-carbon energy solution, this study suggests that it will be feasible to reliably integrate the high levels of zero-carbon energy called for by the Transition Scenario. Achieving this future will require only incremental improvements in technology and operational practices, including continuation of the current trend toward better interregional coordination and intermittent resource capacity forecasting.
Our findings are consistent with other studies, such as MIT 2012, which suggest that much of the U.S. grid could integrate and balance many times the current level of renewables with no additional reliability issues. Recent improvements in both renewable technologies themselves and in the technologies that are used to control and balance the grid have been proceeding at a rapid pace, and the incentives and rewards for success in this area continue to drive substantial progress. In contrast, the alternative—continuing to rely on increasing combustion of fossil fuels to generate electricity, and producing ever-increasing levels of greenhouse gases—is far less feasible, and presents much more daunting technical, economic, and social challenges to human and environmental welfare. In comparison, the challenge of integrating increasing levels of solar and wind power on the U.S. power grids requires only incremental improvements in technology and operational practices.