By Earl J. Ritchie, Lecturer, Department of Construction Management
The problem with variable renewable energy (VRE) – primarily wind and solar – is sometimes it generates too much power and sometimes it doesn’t generate enough. That’s manageable, but it’s more complicated than it may seem.
In the majority of today’s installations, variability can be balanced with so-called dispatchable generation: traditional power plants, hydroelectric and biomass. Generation from traditional power plants is cut when generation from wind and solar is too high, and increased when it’s too low. This creates some power management problems but is manageable at modest cost.
In a system with a large share of wind and solar, maintaining enough dispatchable power in reserve becomes expensive. The electrical grid must be modified to manage the increased variability. It remains to be seen how quickly the transformation to a high share of VRE can be made.
The nature of variability
Power from wind and solar varies on all time scales from seconds to years. The graph below illustrates variation in Irish wind power over one year. The Irish example is pertinent because at 23% of electricity generated, they have one of the highest shares of wind power, and the wind farms are dispersed over the country. Despite the benefit of the geographic spread, there are moderately long periods during which little or no electricity was generated by wind. The historical average amount generated is 31% of installed capacity according to EirGrid and SONI, but the range is from near zero to about 50%.
Managing variability is not new
Variability is not a new issue in the power industry since traditional power sources have some variability, and demand is also variable over all timeframes. The graph below of demand in a large U.S. grid has much less variability than the Irish wind power example, but it is still almost 3:1 and has a noticeable seasonal component.
Managing the system is a function not only of source variation, but also of matching generation with demand. In an earlier post I discussed the “duck curve” illustrating the ramp down and ramp up needed in dispatchable generation due to the mismatch of daily solar generation peaks with demand.
Reducing source variability
Variability can be reduced by combining different types of variable sources and by spreading sources over a large geographic area. Either of these will reduce short-term variability but may or may not significantly reduce variability on a scale of hours or days.
A study of the European Union showed that wind power in 2014 fell to as low as 4% of capacity and was less than 10% of capacity 11% of the time, even when aggregated over the entire EU. Since the countries are not all grid connected, the distribution was hypothetical. Variation on the actual smaller grids was higher.
Patterns of available wind and solar power vary tremendously with location. Wind and solar may tend to peak together or at different times. They may generate more during peak demand periods or during low demand periods. This makes generation design a local issue unless very widespread interconnections are available.
The potential for greater smoothing has led to the concept of the supergrid, connecting generating sources over larger areas than traditional grids. Some technological development is necessary to implement supergrids but they likely will be constructed. Even so, they will not completely eliminate variability since weather patterns tend to occur over large areas.
Reducing demand variability
Variability of demand can be reduced by a variety of techniques that shift usage from high demand periods. These include differential pricing, smart controls, jawboning and direct utility control of load. Perhaps the most obvious example is encouraging people to shift tasks such as washing and drying to the night in order to reduce demand during the daytime peak. These methods are discussed within the industry along with methods for reducing overall demand under the term demand-side management.
Managing the remaining variability
In existing grids and those foreseeable in the near term, substantial variability and mismatch between generation and demand will continue. Management methods include dispatchable generation, overcapacity, storage and tolerating insufficiency. All have costs.
Dispatchable generation is the traditional method. In effect, it is a form of overcapacity since the dispatchable plants run below capacity until more electricity is needed. The cost of maintaining standby capacity and efficiency losses associated with ramping and partial load operation can be substantial.
Renewables can serve as dispatchable sources, so this method would not preclude achieving 100% renewables. Some very high renewables scenarios use biomass to balance variability.
The premise of overcapacity is that if you build more generation than is necessary, you will have enough even when the variable sources operate at a fraction of their capacity. As the graph of Irish wind power shows, it is a practical and economic impossibility to build enough variable capacity to meet supply during very low periods.
The downside of overcapacity is that you generate too much electricity during favorable periods of high wind or intense sunlight. Ideally, the excess electricity can be stored. This has some disadvantages which will be discussed below.
A possibility suggested by Mark Jacobson and Mark DeLucchi is generating hydrogen during periods of oversupply. In essence, this is increasing demand to match supply, and it could be applied to products other than hydrogen. It is conceptually similar to encouraging electricity use by very low or negative prices during oversupply periods, as has been practiced in Germany and other areas with moderately high VRE share.
Storage to clip the peaks and fill the valleys of demand is part of nearly all high VRE scenarios. There are numerous storage technologies with varying cost, scale, duration and technological maturity. This table from Lazard’s 2016 Levelized Cost of Storage shows the cost of the primary technologies and applications. The costs should be taken only as approximations since some of the technologies are not mature, costs vary with location and future cost reductions are likely. Taken at face value, only compressed air, pumped hydro and lithium-ion are competitive today with natural gas peaking cost of about $200 per megawatt hour.
Storage cost depends not only on the cost per kilowatt hour, but also on the amount of storage capacity installed. There are no guidelines for the amount of storage needed for a given level of VRE. The optimum capacity is influenced by cost dependent tradeoffs between generation and storage, as well as the mix of sources and match with demand. A model study of the PJM Interconnection used as the demand example above showed the lowest cost alternative relied heavily on overcapacity, with little storage. Other locations and assumptions might give very different answers.
Storage technology is in an early stage of development. Most storage installations to date can only supply rated power for a few minutes to a few hours. Capability to handle extended shortage remains an issue. The extent of storage that will be incorporated in future systems will be heavily dependent upon development of storage methods and cost of generation.
It is likely impossible to build a grid with a very high share of VRE that has complete certainty of providing adequate power at all times. A necessity or deliberate choice may be to allow for curtailment, that is, not supplying some customers when generation does not equal demand.
Market mechanisms, such as interruptible supply contracts, are other ways to match supply and demand.
Optimizing the system
On a theoretical basis, an electrical grid can be optimized through the proper mix of sources, storage and locations. There is a question what is to be optimized. Is it lowest cost, least pollution, greatest economic benefit, energy security, social equity or some combination of factors? Once the measure is determined, assumptions must still be made regarding performance, cost and demand. Actual performance will frequently differ from modeled performance.
Since the amount of electricity generated by wind and solar vary somewhat randomly, statistical forecasting techniques are used. These generate a distribution of forecasted supply as a function of time. There will be some probability of extreme events, for example, a prolonged inadequacy of supply.
The choice of when, where, how much and what type of generation to build is decided in most countries by private companies. Their choices may be substantially influenced, but not controlled by, government policy. As a result, the grid will not be optimum. Renewables requirements and the structure of government incentives will be important factors.
Very high renewables scenarios
The majority of published scenarios, including those of the IPCC, have traditional sources – nuclear and fossil fuels – continuing to provide a significant fraction of electricity generation through 2050. A few have all electricity, or even all primary energy, from renewables. These scenarios depend not only on rapid technological advancement and implementation of renewable sources, but also on reduction of energy consumption, such as in this World Wildlife Fund (WWF) scenario of 95% renewables.
Source: World Wildlife Fund
The WWF scenario decreases overall energy demand by about 25% from a peak in 2020. It is at odds with many other scenarios that envision continued growth in energy demand due to increasing population and increases in consumption in the developing and less developed countries.
Similarly, this scenario envisions a decrease in annual energy cost of 4 trillion Euros by 2050, based on reduced demand and lower fuel costs. These numbers are at odds with the predicted increase in generation cost associated with high shares of VRE discussed in an earlier post.
It’s not clear to me whether scenarios that envision drastic shifts in energy source are considered plausible or are thought experiments expressing ideal goals. The WWF report describes the task of transforming the system as “a huge one, raising major challenges.” Considering the modest progress to date, differing views of the priority of decarbonization, the need for as yet unproven technology and the time needed to construct new systems, it seems unlikely that this transformation will be completed by 2050.
Earl J. Ritchie is a retired energy executive and teaches a course on the oil and gas industry at the University of Houston. He has 35 years’ experience in the industry. He started as a geophysicist with Mobil Oil and subsequently worked in a variety of management and technical positions with several independent exploration and production companies. Ritchie retired as Vice President and General Manager of the offshore division of EOG Resources in 2007. Prior to his experience in the oil industry, he served at the US Air Force Special Weapons Center, providing geologic and geophysical support to nuclear research activities.