The benefits and challenges of large scale PV penetration have been described in a number of analyses (Brinkman et al 2011). At low penetration, PV typically displaces the highest cost generation sources (Denholm et al.
2009) and may also provide high levels of reliable capacity to the system (Perez et al 2008). Figure 1 provides a simulated system dispatch for a single summer day in California with PV penetration levels from 0% to 10% (on an annual basis). This figure is from a previous analysis that used
a production cost model simulating the western United States (Denholm et al. 2008). It illustrates how PV displaces the highest cost generation, and reduces the need for peaking capacity due to its coincidence with demand patterns.
At fairly low penetration (on an energy basis) the value of PV capacity drops. This can be observed in Figure 1 where the peak net load (normal load minus PV) stays the same between the 6% and 10% penetration curves. The net load in this figure is the curve at the top of the “Gas Turbine” area. Beyond this point PV no longer adds significant amounts of firm capacity to the system. Several additional challenges for the economic deployment of solar PV also occur as penetration increases. These are illustrated in Figure 2, which shows the results of the same simulation, except on a spring day. During this day, the lower demand results in PV displacing lower cost baseload energy. At 10% PV penetration in this simulation, PV completely eliminates net imports, and California actually exports energy to neighboring states.
§ 10,000 (3
Several factors limit the ability of conventional generators to reduce output to accommodate renewable generation. These include the rate at which generators can change output, particularly in the evening when generators must increase output rapidly in a high PV scenario. This challenge is illustrated in Figure 3, a ramp duration curve for California covering an entire simulated year. This is the net load ramp rate (MW/hour) for all 8,760 hours in the simulated year ordered from high to low. In the no PV case, the maximum load ramp rate is about 5,000 MW/hour and a ramp rate of greater than 4,000 MW/hour occurs less than 100 hours in the simulated year. In the 2% PV case, the hourly ramps are actually smaller since PV effectively removes the peak demand (as seen in Figure 1). However at higher penetration, the ramp rates increase substantially, and in the 10% PV case the net load increases at more than 4,000 MW/hour more than 500 hours per year.
Another limitation is the overall ramp range, or generator turn-down ratio. This represents the ability of power plants to reduce output, which
is typically limited on large coal and nuclear units. Accommodating all of the solar generation as shown in Figure 2 requires nuclear generators to vary output which is not current practice in the U. S. nuclear industry. Most large thermal power plants cannot be turned off for short periods of time (a few hours or less), and brief shutdowns could be required to accommodate all energy generated during the period of peak solar output. The actual minimum load of individual generators is both a technical and economic issue—there are technical limits to how much power plants of all types can be turned down. Large coal plants are often restricted to operating in the range of 50%-100% of full capacity, but there is significant uncertainty about this limit (GE Energy 2010). Many plant operators have limited experience with cycling large coal plants, and extensive cycling could significantly increase maintenance requirements.
The ability to “de-commit” or turn off power plants may also be limited by the need to provide operating reserves from partially loaded power plants. As the amount of PV on the system increases, the need for operating reserves also increases due to the uncertainty of the solar resource, as well as its variability over multiple time scales.
Previous analysis has demonstrated the economic limits of PV penetration due to generator turn-down limits and supply/demand coincidence (Denholm and Margolis 2007a, Nikolakakis and Fthenakis 2011). Because of these factors, at high penetration of solar, increasing amounts of solar may need to be curtailed when its supply exceeds demand, after subtracting the amount of generation met by plants unable to economically reduce output due to ramp rate or range constraints or while providing operating reserves. Generator constraints would likely prevent the use of all PV generation in Figure 2. Nuclear plant operators would be unlikely to reduce output for this short period. Furthermore, PV generation may be offsetting other low or zero carbon sources. In Figure 2, PV sometimes displaces wind and geothermal generation, which provides no real benefit in terms of avoided fuel use or emissions.
While the penetration of solar energy is currently far too small to see significant impacts, curtailment of wind energy is an increasing concern in the United States (Wiser and Bolinger 2010). While a majority of wind curtailments in the United States are due to transmission limitations (Fink et al 2009), curtailments due to excess generation during times of low net load are a significant factor that will increase if grid flexibility is not enhanced. The resulting curtailed energy can substantially increase the levelized cost of energy (LCOE) from variable generators, because their capital costs must be recovered over fewer units of energy actually sold to the grid.
The ability of the aggregated set of generators to rapidly change output at a high rate and over a large range can be described as a grid’s overall flexibility. Flexibility depends on many factors, including:
• Generator mix:Hydro and gas-fired generators are generally more flexible than coal or nuclear.
• Grid size: Larger grids are typically more flexible because they share a larger mix of generators and can share operating reserves and a potentially more spatially diverse set of renewable resources.
• Use of forecasting in unit commitment: Accurate forecasts of the wind and solar resources and load reduces the need for operating reserves.
• Market structure: Some grids allow more rapid exchange of energy and can more efficiently balance supply from variable generators and demand.
• Other sources of grid flexibility: Some locations have access to demand response, which can provide an alternative to partially-loaded thermal generators for provision of operating reserves. Other locations may have storage assets such as pumped hydro.
A comprehensive analysis of each flexibility option is needed to evaluate the cost-optimal approach of enhancing the use of variable generation. In this analysis, we consider the use of thermal energy storage. Previous analysis has demonstrated the ability of a wind and solar-based system to meet a large fraction of system demand when using electricity storage (Denholm and Hand 2011). A number of storage technologies are currently available or under development, but face a number of barriers to deployment including high capital costs efficiency related losses, and certain market and regulatory challenges. A number of initiatives are focused on reducing these barriers.
An alternative to storing solar generated electricity is storing solar thermal energy via CSP/TES. Because TES can only store energy from thermal generators such as CSP, it cannot be directly compared to other electricity storage options, which can charge from any source. However, TES provides some potential advantages for bulk energy storage. First, TES offers a significant efficiency advantage, with an estimated round trip efficiency in excess of 95% (Medrano et al. 2010). TES has the potential for low cost, with one estimate for the cost associated with TES added to a CSP power tower design at about $72/kWh-e (after considering the thermal efficiency of the power block).