Intermittent energy source

Intermittent power sources are sources of power generation, primarily electricity, whose power output is either variable or intermittent. In contrast to other sources of energy, intermittent power sources are generally not dispatchable, that is, the ability of an operator to increase output to meet power demand is limited. Many sources of renewable energy are considered intermittent, such as wind and solar energy. Since demand and supply of electricity must be balanced at all times to maintain grid stability, the variable output of intermittent power sources creates significant challenges for grid operators, and can substantially affect the economics of power generation using such sources.

Solutions for managing mismatches between demand and supply exist in all managed grids, and include supply management (increasing or decreasing energy output from grid-connected plants) and demand management (increasing or decreasing demand, which may include load shedding, demand management or energy storage for later use). Many grids also use energy pricing to influence supply and demand (increasing prices to encourage increased supply and lower demand), but current pricing solutions are incomplete due to different time frames needed to find "market pricing" solutions and ensure stability of the grid. All grids are, therefore, regulated to some degree.

Strictly speaking, all sources of electric power generation can suffer from intermittency and variability. Even fully dispatchable energy plants may go off-line for various technical reasons, and their energy production may differ from that demanded or predicted, although by construction, dispatchable sources are by definition those that have a high degree of predictability and capacity to meet output targets within certain limits.

Intermittency is most properly used to refer to power output that may go off-line entirely at various times: that is, the power output states have a binary or on-off nature. Variable power sources may show substantial differences in output, but generally would not "trip" on or off in extremely short periods of time (particularly with greater geographical spread or distribution). Throughout this article, the two terms will generally be used interchangeably, as intermittency has become the most commonly used term to describe this issue.

This article primarily concentrates on sources of electricity generation that are not generally considered dispatchable, that is, whose energy output are primarily dependent on exogenous elements beyond the control of operators.

Intermittency and Renewable Energy

Intermittency strongly affects many sources of renewable energy, as they are dependent on natural processes that are to some degree unpredictable. The timeframe under consideration strongly affects the degree to which the problem is considered relevant, however. For example, biomass renewables are dependent on solar energy and weather conditions, but the timeframe is substantially longer; in addition, biomass itself represents a form of energy storage, and the decision of when to utilize the stored energy is to some degree controllable.

Two forms of intermittent renewable energy, wind and solar electricity generation, present the most significant challenges due to the timeframe of changes in generation and the limited correlation with demand cycles. Both can vary greatly throughout the course of a diurnal cycle, and may also be subject to substantial seasonal variation. In addition, day-to-day power generation may vary significantly (due to prevailing winds or cloud cover) with limited predictability. Similarly, the ability of operators to control output for both is generally limited to curtailment: power output can be decreased, but generally not increased at will. Curtailment of output is a common feature in electrical grids and for wind and solar installations, but reducing power sold to the grid may substantially affect project economics.

Hydropower can be both intermittent or dispatchable, depending on the configuration of physical plant. Typical hydroelectric plants in the dam configuration may have substantial storage capacity, and be considered dispatchable. Run of the river hydroelectric generation will typically have limited or no storage capacity, and be intermittent on a seasonal or annual basis (dependent on rainfall and other factors). Hydroelectric dams have limits to dispatchability, since storage is finite and there are often environmental and regulatory requirements that detail minimum and maximum release into the water system. Hydrostorage is also used in many locations to manage supply and demand, but true hydrostorage is not an energy source: it is a mechanism used to store excess generating capacity for use during times of greater demand.

Intermittency: Wind Energy

Main article: Wind power

Cement works in New South Wales, Australia. Energy-intensive process like this could utilize burst electricity from wind.

Wind-generated power is a variable resource, and the amount of wind-generated electricity produced at any given point in time by a given plant will depend on wind speeds and turbine characteristics (among other things). As compared to many other types of electricity generation, wind is not dispatchable - it generally cannot be turned on or off at will by human or automatic dispatch. Variability may be a more accurate term to describe wind's generation profile than intermittency, which may imply an alternating presence or absence (generation that is either on or off). In discussions of the pros and cons of wind power, the issue of variable power output may be termed intermittency or variability without distinction between the two terms.

As wind energy installations grow in absolute terms and as a proportion of existing output, concerns have been raised about integrating wind energy in existing grids due to the variability of power output from wind. In combination with the output profile of existing power plants, concern has been expressed about the extent to which wind can be relied upon for output during periods of high demand, sometimes referred to as its base capacity factor. This is distinct from the annual capacity factor, which is the average output a year divided by the turbine's nameplate capacity. The base capacity factor is primarily a function of the statistically reliable output of wind during the period of peak demand in a given area, as these are the times when the capacity to meet demand without threatening grid stability or shortages will be most critical.

The economics of wind energy may also be challenged when wind production is high at times of low demand. Due to the presence of other generating stations that are operated as base load (run as close to continuously as possible) or have minimum operating cycles, at high penetrations wind plants may contribute to the grid producing energy "surplus" to requirements at times of low demand. As with other generating plant, wind energy output may on occasion need to be curtailed, energy stored for later use, or load increased to compensate. While all of these solutions are commonly used to manage grids, wind "spilt" or curtailed generates no revenue, and prices for supply to the grid may be lower at times of high output, both of which could make wind farms less profitable. Energy storage used to arbitrage between periods of low and high demand may be capital intensive and always incurs some efficiency losses.

Both shortfalls and surpluses of supply attributable to wind energy's variability will be less frequent at lower penetration levels. At low to medium levels of penetration (up to 15%), incremental regulation and operational reserve requirements are generally marginal, and demands for reduced supply infrequent. At lower levels (less than 5%), wind may be treated as "negative load" in the larger system. Very few grids have wind energy penetration above these levels.

Penetration

• As the fraction of energy produced by wind ("penetration") increases, different technical and economic factors affect the need for grid energy storage facilities, demand side management, and/or other management of system load. Large networks, connected to multiple wind plants at widely separated geographic locations, may accept a higher penetration of wind than small networks or those without storage systems or economical methods of compensating for the variability of wind. In systems with significant amounts of existing pumped storage, hydropower or other peaking power plants, such as natural gas-fired power plants, this proportion may be higher. Isolated, relatively small systems with only a few wind plants may only be stable and economic with a lower fraction of wind energy (e.g. Ireland).
• Depending on the profile of other generation, strong wind generation at times of low demand may result in an excess of supply, which can harm grid stability, as certain generation types are not maneuverable. If mechanisms to export, store or otherwise divert this energy do not exist or are insufficient, wind turbines may have to curtail their output (for example, by changing the pitch of the turbine blades). This is a normal operating procedure that can be handled by turbine operators and control software. It reduces, however, the revenue generated by the wind plant and will affect the economic viability of wind production. Some jurisdictions - notably Germany - require grid operators to purchase from renewable sources first. In other cases, grid pricing procedures may allow for nil or negative prices, providing incentives to market participants to curtail production or increase load (for example, for storage). Excess supply events which may require curtailment can be expected to increase with wind penetration, which may also encourage the development of storage solutions.
• Although penetration is generally stated in terms of nameplate capacity (peak output) over peak demand, at higher penetrations of wind generation, penetration may be better measured as peak wind output over low demand plus export and storage may be more relevant[1]. Variations on this approach may more accurately capture the likelihood of "excess" supply during periods of high wind output, and the ability of the system to economically absorb additional wind. There remain no generally-accepted levels of maximum penetration, however.
• On most large power systems a moderate proportion of wind generation can be connected without the need for storage. For larger proportions, storage may be economically attractive or even technically necessary. The profile of other generation facilities in the system (nuclear, coal, natural gas, hydro, etc.) will also influence the potential need for storage. At present, there are few large systems (for example, at the national or regional level) with sufficiently high wind generation to drive demand for storage (or other solutions, such as export, have been more economical), and discussion of the issue and potential upper limits for wind penetration remain largely hypothetical.
• Electricity demand is variable but generally very predictable on larger grids; errors in demand forecasting are typically no more than 2% in the minutes-hours-day ahead timeframe. Depending on the demand profile and location, local weather conditions - particularly temperature - may be the primary driver of demand, and the sensitivity of demand to prediction errors may be well understood. Wind energy production can also be forecast, but there is considerably less experience predicting wind speeds, and the time frame of forecasts and sensitivity factors less well understood. At present, error rates for predicting wind production at important timeframes for grid operators (hours and day-ahead) are significantly higher than for predicting demand.
• The maximum proportion of wind power allowable in a power system will thus depend on many factors, including the size of the system, the attainable geographical diversity of wind, the conventional plant mix (coal, gas, nuclear, hydroelectric) and seasonal load factors (heating in winter, air-conditioning in summer) and their statistical correlation with wind output. For most large systems the allowable penetration fraction (wind nameplate rating divided by system peak demand) is thus at least 15% without the need for any energy storage whatsoever. Note that the interconnected electrical system may be much larger than the particular country or state (e.g. Denmark, California) being considered. A study published in October, 2006, by the Ontario Independent Electric System Operator (IESO) found that "there would be minimal system operation impacts for levels of wind capacity up to 5,000 megawatts (MW)," which corresponds to 17% of projected peak load (nameplate wind capacity over peak load); at the time of publication, Ontario had only 300 MW of installed wind capacity. While there are both practical and theoretical upper limits (as with any type of electric power generation), these upper limits are frequently many times higher than existing installed capacity.
• The variability of wind can raise costs for regulation and incremental operational reserve. At high penetrations of wind (greater than 30%, depending on a number of factors), storage may become necessary and/or demand management employed; additional storage would likely raise costs for the additional wind energy capacity. Effective demand management would lower the need for peaking power during demand spikes, and reduce the reserve required.
• Wind power generation tends to be higher in the winter and at night (due to higher air density), so the appropriateness of wind power in high concentrations may crucially depend on the prevalence of air conditioning in a given jurisdiction. Wind power may be weakest in the hot summer months, and particularly during the day when air conditioning demand is highest. Conversely, systems where heat is electrical may be well-suited to higher penetration of wind power. Daily generation profiles may vary substantially in different locations.

Geographic Diversity

The variability of production from a single wind turbine can be high. Combining any additional number of turbines (for example, in a wind farm) results in lower statistical variation, as long as the correlation between the output of each turbine is imperfect, and the correlations are always imperfect due to the distance between each turbine. Similarly, geographically distant wind turbines or wind farms have lower correlations, reducing overall variability considerably.

• Paradoxically, whilst wind power is variable, it is highly reliable in the sense that single-unit failure (and the associated loss of generation capacity) is unlikely, and the reliability increases with penetration. For example to power the entire UK National Grid would require about 35,000 x 3MW turbines. If at a given instance they were to be producing in aggregate 60 GW, then it is not possible for all 35,000 turbines to fail simultaneously, nor can the aggregate output deviate at a very great rate, simply because due to the geographic dispersal, wind regimes cannot change instantaneously over a large area. Conversely it is not unheard of for even reliable dispatchable plants to suddenly fail. This means that conventional grids have to have instantaneous reserve equal to the possible failure of two of the largest generating units.
• Multiple wind farms spread over a wide geographic area and gridded together produce power much more constantly and with less variability than smaller installations. Wind output can be predicted with a fair degree of confidence many hours ahead using weather forecasts, especially from large numbers of turbines/farms. The ability to predict wind output is expected to increase over time as data is collected, especially from newer facilities.

Compensating for Variability

As noted, all sources of electrical power have some degree of unpredictability, and demand patterns (while relatively predictable) routinely drive large swings in the amount of electricity that suppliers feed into the grid. Wherever possible, grid operations procedures are designed to match supply with demand at high levels, and the tools to influence supply and demand in response to conditions are well-developed. The introduction of a power source in large amounts, however, that has a random element, relatively large differences between peak and trough output, and that may not be well-matched to demand cycles may require substantial changes to existing procedures and additional investments.

Operational Reserve & Peak Demand Reduction

At times of high or increasing demand where wind's output may simultaneously be falling, a number of solutions are either commonly used today or potentially feasible.

• Because conventional powerplants can drop off the grid within a few seconds, for example due to equipment failures, in most systems the output of some coal or gas powerplants is intentionally part-loaded to follow demand and to replace rapidly lost generation. The ability to follow demand (by maintaining constant frequency) is termed "response." The ability to quickly replace lost generation, typically within timescales of 30 seconds to 30 minutes, is termed "spinning reserve." Nuclear power plants in contrast are not very flexible and are not intentionally part-loaded. A power plant that operates in a steady fashion, usually for many days continuously, is termed a "base load" plant. Generally thermal plants running as "peaking" plants will be less efficient than if they were running as base load. Hydroelectric facilities with storage capacity (such as the traditional dam configuration) may be operated as base load or peaking plants, and complement high levels of wind penetration.
• What happens in practice therefore is that as the power output from wind varies, part-loaded conventional plants, which must be there anyway to provide response (due to continuously changing demand) and reserve, adjust their output to compensate; they do this in response to small changes in the frequency (nominally 50 or 60 Hz) of the grid. In this sense wind acts like "negative" load or demand.
• Energy Demand Management or Demand-Side Management refers to the use of communication and switching devices which can release deferrable loads quickly to correct supply/demand imbalances. Incentives can be created for the use of these systems, such as favorable rates or capital cost assistance, encouraging consumers with large loads to take advantage of renewable energy by adjusting their loads to coincide with resource availability. For example, pumping water to pressurize municipal water systems is an electricity intensive application that can be performed when electricity is available. Real-time variable electricity pricing can encourage all users to reduce usage when the renewable sources happen to be at low production.

Storage and Demand Loading

At times of low or falling demand where wind's output may be high or increasing, grid stability may require lowering the output of various generating sources or even increasing demand, possibly by using energy storage to time-shift output to times of higher demand. Such mechanisms can include:

• Long-term storage of electrical energy involves substantial capital costs, space for storage facilities, and some portion of the stored power will be lost during conversion and transmission. The percentage retrievable from stored power is called the "efficiency of storage." The cost of compensating for the variability of wind has been studied extensively at low to medium penetrations, but would be expected to rise with higher penetration levels; the increase in costs with significantly higher penetration may be non-linear as the variability becomes more significant at higher levels, and particularly if storage needs to be purpose-built for wind.See: Grid energy storage
• The allowable penetration may be further increased by increasing the amount of part-loaded generation available, or by using energy storage facilities, although if purpose-built for wind energy these may significantly increase the overall cost of wind power. Systems with existing high levels of hydroelectric generation may be able to incorporate substantial amounts of wind, although high hydro penetration may indicate that hydro is already a low-cost source of electricity; Norway, Quebec, and Manitoba all have high levels of existing hydroelectric generation (Quebec produces over 90% of its electricity from hydropower). Storage capacity in hydropower facilities will be limited by size of reservoir, and environmental and other considerations.
• Existing European hydroelectric power plants can store enough energy to supply one month's worth of European electricity consumption. Improvement of the international grid would allow using this in the relatively short term at low cost, as a matching variable complementary source to wind power. Excess wind power could even be used to pump water up into collection basins for later use. In practice, Denmark's system is well-integrated with the hydro-electric dominated Norwegian system, and Norwegian hydropower is used to balance fluctuations and shortfalls in Denmark; on occasion, Denmark exports electricity to Norway when generation is higher than demand (thereby increasing stored hydropower). Increased wind penetration may raise the value of existing peaking or storage facilities and particularly hydroelectric plants, as their ability to compensate for wind's variability will be under greater demand.
• In energy schemes with a high penetration of wind energy, secondary loads, such as desalination plants and electric boilers, may be encouraged because their output (water and heat) can be stored. The utilization of "burst electricity", where excess electricity is used on windy days for opportunistic purposes greatly improves the economic efficiency of wind turbine schemes. An ice storage device has been invented which allows cooling energy to be consumed during resource availability, and dispatched as air conditioning during peak hours.

Complementary Power Sources and Matching Demand

• Electricity produced from solar energy could be a counter balance to the fluctuating supplies generated from wind. In some locations, it tends to be windier at night and during cloudy or stormy weather, so there is likely to be more sunshine when there is less wind.
• In some locations, electricity demand may have a high correlation with wind's output, particularly in locations where cold temperatures drive electric consumption (as cold air is denser and carries more energy).

Maximum Penetration Limits

However, there is no overwhelming technical or economic reason why interlinked national grids could not have close to 100% of annual delivered energy generated from wind turbines for probably around 10% extra on the cost of a present domestic unit.

This is because:

1. The source of power generation for the few hours or days every year of no wind generation, can be provided by retaining the existing power stations which are already built and largely paid for, and starting them up as required. They can also be used to "fill in" missing power when load exceeds the aggregate windfarms output. The cost of using existing power stations in this way is surprisingly low since 96% of running a power station is the fuel cost. The actual cost of paying to keep a power station idle, but useable at short notice, can be readily calculated from published spark spreads and dark spreads.

2. The aggregate maximum rate of change of generation from a close to 100% wind scenario,in the UK (which would necessarily be comprised of geographically dispersed wind farms), is well under the already existing rate of change of total generation due to unscheduled power station outages, or the unexpected rate of change of demand that existing large grid and power station systems are presently already required to routinely deal with. Such maximum rate of change in the UK is typically the simultaneous loss of two of the largest generating units on the system i.e., 2 x 660 MW = 1.32 GW, in fact Sizewell B( with additional reserve stations to cover the the closure due to a type fault of 3 more units). This maximum rate of sudden change always exceeds the rate of change of any sudden load pick up at eg the end of a popular tv program. These load pick ups can be bigger - 3 GW in overall magnitude but happen more slowly.

• http://www.eci.ox.ac.uk/publications/downloads/sinden05-dtiwindreport.pdf for detailed study of UK wind regime and its actual variability for geographically separated wind farms looked at over 30 years.

3. The aggregate rate of change of output of such a close to 100% wind scenario is also much slower than the rate at which power stations can be warmed / started and and ramped up, and successive short term weather forecasts (from 12 hours for an initial forecast to 5 minutes for final balancing are sufficiently good to have the right amount of available but not running plant, warming plant and spinning reserve plant always available to cover for any likely sudden change in aggregate wind generation. The bigger the interconnected grid - say going from UK only to European level the more this slowness of rate of change applies. The constant 12 hour forecasting and short term (5 minute, 1 minute) adjustment is no more than what happens at present in the control room of in any interconnected grid operator.

4. Automatic load shedding of large industrial loads and its subsequent automatic reconnection is well established and routine in the UK and US grids at least, and in the UK these are known as Frequency Service contractors. They are under contract and receive a fee proportional to the capacity they disconnect, typically £7k per MW per annum. Several GW are switched off and on each month in the UK in this way.

Operating hand in hand with the above Frequency Service contractors are Reserve Service contractors who offer fast reponse gas turbines and even faster diesels again this is also already widely used in the UK, France and US to control grid stability. This combination of Reserve Service and Frequency Service enables much lower levels of the more expensive spinning reserve (ie innefficient part loaded stations) to cope with the largest sources of existing intermittency which is the power stations themselves, and which intermittency would still remain the largest even in a close to 100% wind scenario. To achieve this effect the load shedding and diesels only have to be operated a few hours each month.

These systems can be readily and cheaply extended although in the UK it is hard to see why they would need much extension. (because the rates of change caused by power station failure at present exceed anything a 100% wind scenario would imply)

In a close to 100% wind scenario, surplus wind power can be catered for by increasing the levels of the existing Reserve and Frequency Service schemes whereby a rise in system frequency would automaticaly connect loads, and disconnect them later, and by extending the scheme to more domestic sized loads. This means energy can either be stored by advancing deferable domestic loads such as storage heaters, water heaters, fridge motors, or even hydrogen production. Just as automatic systems already exist to achieve this at an industrial level, they also exist at a domestic level.

Alternatively or additionally, power can be exported to neighboring grids and re-imported later. EHVDC cables are now extremely cheap and efficient (3% loss per 1000km)- Holland is just building one 500 km link to connect it to Norway. The longest presently is 1500 km. Central Germany has been connected to Norway's hydro since the mid 70's. UK is connected to Europe via a 2 GW link and to Ireland. There are numerous well advanced plans for other inter-continental connections of several thousand km.

It can be shown that the increased cost of a close to 100% wind scenario covering increased back up for windless days, reserve, transmission and balancing costs is likely to be about an extra 10% on the price of a delivered unit to the customer.

http://en.wikipedia.org/wiki/National_Grid_UK#Triads How TRIADS can be used to estimate the increase in costs for installing close 100% wind scenario.

http://en.wikipedia.org/wiki/Spark_spread#Spark_Spread_as_cost_of_back_up Wikipedia for how to calculated the cost of back up from published Spark and Dark Spreads.

http://en.wikipedia.org/wiki/How_the_UK_National_Grid_is_presently_controlled Wikipedia article for how large national grids are actually controlled, in detail, and which estimates the costs mentioned above.

This 10% figure ignores the potential cost of the domestic scale Reserve and Frequency Service extesion and any transnational cables

There will be other costs, over and above the 10% estimate presented here, associated with short term storage (such as switchable off peak space and water heaters) and export and re-import costs via inter-country cables, but since sufficient of these methods were provided to cover the increase in nuclear in the UK in the 60s (30% of the then total capacity) i.e., the construction of the 2 GW Dinorwig and FFestiniog pumped storage, the introduction of 5m homes with off peak heating it is hard to see why these costs should impact significantly on the overall economics of 100% wind, anymore than they did with nuclear. This kind of load switching is already widely used in New Zealand to control domestic water heaters to flex load. These type of measures in any case tend to be self financing due to the savings from load smoothing, and the resultant cut in peak capacity and back up of conventional plant.

http://www.enermet.com/en/load_management/ripplecontrol.php

It goes without saying, that there would be no point to replacing existing hydro facilities with wind.

http://eeru.open.ac.uk/conferences.htm#jan06 Power Point presentation showing how national generating systems are actually controlled in detail.