Contrary to popular belief -- at least that held by wind power's detractors, a large body of utility opinion, and now a group of economists -- the introduction of wind energy into an integrated electricity network will not result in substantial additional operating costs. Even with significant amounts of wind on the grid, dedicated back-up generation is not needed. That fact, however, is not being allowed to get in the way of policy makers and economists now charged with devising rules and regulations for competitive electricity markets.
The economic mechanism they have come up with for ensuring that supply matches demand at the macro level is a system of "balancing charges." Generators or suppliers, who fail to match supply and demand will be at the mercy of a "balancing market," a forum where surpluses and deficiencies of power can be bought and sold. A balancing market is at the heart of the New Electricity Trading Arrangements (NETA) now in the process of gestation in the UK and is being run by NordPool in Scandinavia. A similar mechanism is operated by the Automatic Power Exchange in California.
The problem with the balancing charges mechanism -- as it is structured -- is that it makes a mockery of market forces. Most of the time, no extra costs are incurred on an integrated power system by generators who fail to meet their scheduled deliveries. Dispersed variations in both supply and demand are balanced by the aggregation of power in the system as a whole. Non delivery of a few megawatts, or the uptake of an overflow, costs the system nothing.
Generators and suppliers, however, will have to pay balancing charges. These are incurred because they are forced to enter a short term trading market to sell power surpluses (at the risk of making a loss) or buy power to make up a deficit (at the risk of being charged a premium). Real price signals -- the lifeblood of a competitive market -- are hidden behind this "virtual" trading mechanism. It is not the aim of policy makers to distort the market and charge generators for non existent costs, but that will be the result.
False cost signals
The need to match supply with demand on an hourly basis is a major complexity in replacing the old "command and control" approach to electricity supply with the competitive efficiencies of market forces. The fragmenting effects of liberalisation on what until now has been a centrally planned and controlled industry means generators are treated as isolated competitive elements, even though they are actually part of an aggregated whole. Electricity, and that from wind energy in particular, is not a product which lends itself to neatly separated bi-lateral contracts between generators and suppliers -- the kind competitive markets demand. Stockpiling in warehouses is not an option for storing surpluses or fetching extra supplies, so a balancing market is needed to fulfil that role.
A properly functioning balancing market should mirror the workings of a modern electricity system by aggregating all the surpluses and deficits resulting from bi-lateral contracts so that the scales of supply and demand are level. The balance can be fine tuned using the existing reserve: a typical margin for "scheduling error" on a system is 1-2% of demand. The variations of a fluctuating source like wind get absorbed in the process, even when it makes up a significant proportion of generation. The balancing effects of aggregation, however, are not taken into account in Britain's New Electricity Trading Arrangements.
Under NETA all generators are required to submit final output forecasts four hours in advance based on their bi-lateral contracts with suppliers. Unscheduled events, however, such as a surge in demand or a power plant going off-line, require both generators and suppliers to dip into the balancing market to buy power to make up a shortfall, or to sell a surplus. It is a requirement which distorts the real market by sending a false economic signal that suggests there is no aggregation.
Generators who cannot forecast their output with great accuracy are particularly penalised. Wind will have to go to the balancing market to make up a deficit of supply or to sell surpluses far more often than thermal generators because wind strengths can change unpredictably within the space of four hours. The upshot is that wind operators in Britain are now threatened with substantially higher "balancing charges" than the real costs to the system they incur by getting a delivery forecast wrong. The size of this effective penalty is still unknown, but National Wind Power, the country's largest wind plant developer, estimates it will be around £0.004/kWh. As such, sites with wind speeds less than around 7.5 m/s become uneconomic to develop, restricting the national wind resource significantly. Forcing wind to operate in such a market will distort its price and reduce its competitive ability.
In short, the economic theory behind NETA does not match the economic practice of an integrated power system. The UK balancing market is expected to meet anything between 10% and 50% of demand at any one time -- and perhaps even more -- so the mismatch between theory and practice could result in severe distortions of the overall market. Needless to say, wind plant owners and consumers will be the financial victims, while the owners of conventional generating plant will be the winners.
To deduce the true cost implications of operating wind power plant in an integrated electricity system requires an understanding of power system fundamentals. Overall swings in demand are much smoother than those of individual consumers and the same rule applies to generation. Where a typical household may have a 15-to-1 ratio between peak and average demand, the corresponding ratio for an electricity system is typically around 1.5-to-1. Fluctuations from wind become lost in the system in the same way. What matters is the extent to which adding wind to an integrated system adds to the overall uncertainty in the supply and demand balance.
Dedicated "back-up" generation for wind power will only ever be needed if wind was to make up a considerably larger proportion of supply than so far envisioned -- and only if the wind plant were concentrated in a small geographical area. This is not just starry eyed theory. Wind energy penetration in large areas of Denmark, Germany and even Ireland is proving the point in practice. Even in extreme conditions, such as the onset of a hurricane in Denmark when large numbers of wind turbines temporarily stopped production, power outages are not the result of too much or too little wind.
The task of demonstrating that wind is not a threat has become easier with the publication of two unrelated studies in the past year. The first is an assessment by Britain's National Grid Company (NGC) of the criteria for integration of renewable energy; the second is a set of measurements from Germany's federal R&D program for wind. The combined evidence from both studies shows that wind causes even fewer problems on an integrated system than supposed so far.
NGC, the operator of the British power system, has suggested a series of basic criteria which should govern assimilation of renewables (table). Its approach to wind on a power system can be presumed to be conservative, since security of supply will have the highest priority. Furthermore, the characteristics of electricity generating plant are similar the world over and these criteria will be relevant to other system operators.
NGC's limit for fluctuating generation is in the zone of 20% of peak demand, confirming what has been known for some time. Up to that point a change in output can be absorbed within the system using the normal reserves. At more than 20% penetration, a quiet sunny day would require the purchase of additional controllable output from the market, according to NGC.
Instantaneous loss of generation is more problematic, though if renewables were supplying no more than 2% of maximum demand when they dropped off the grid, the NGC system can absorb the blip. Above 2% and NGC needs to buy additional frequency control measures. In the UK, 2% penetration corresponds to 1000 MW. But 1000 MW of wind would not be concentrated at a single location and it is not plausible to suggest that this amount of distributed wind would all switch off at once.
Denmark -- with the highest concentration of wind anywhere in the world -- proves this point. It has 1750 MW of wind, whereas 2% of maximum demand corresponds to 165 MW -- yet this amount is not known for dropping off the grid all at once. In practice the Danish power system copes with a penetration level for wind of 17% of the total generating capacity of 10,000 MW, even when much of this is not controllable. (The prevalence of combined heat and power plant means that control of electricity output is subject to the demand for heat.)
The third of NGC's criteria -- a 3% penetration threshold for unpredictable short notice loss of output -- is perhaps the most testing. A change in wind's output of 3% of maximum demand within an hour is acceptable. Above that level, however, and the system operator is faced with buying additional reserve services, argues NGC.
In Britain, 3% of maximum demand corresponds to 1500 MW and in Denmark to 250 MW, or 14% of the capacity of the wind plant. In practice, however, even this proportion of wind output hardly ever disappears from the grid within the space of an hour, as recent data from Germany reveals. The probability of a 14% change in an hour is about 0.02%. The maximum change ever recorded within an hour in a capacity base of 350 MW of distributed wind in Germany was about 20% of the capacity of the wind plant, or about 70 MW (Figure 1). In an integrated system, a utility would cope with these changes by measures such as increasing the output of spinning reserve, demand management, or imperceptible changes in voltage.
Estimating the likely behaviour of 1700 MW of wind in Denmark from measurements of 350 MW in Germany is not easy. It is safe to assume, though, that the maximum percentage change will become less as the capacity gets higher, since the diversity of wind turbine siting will increase. In any case, wind capacity in Denmark has already reached a level where coping with fluctuations in wind output is a day-to-day part of utility operating procedures.
Maximum changes in wind output are only part of the story. More gentle fluctuations also have operational implications associated with increasing amounts of wind energy. Significantly, the latest German data is allowing a recalibration of the assessments of fluctuations done to date using statistical methods. This recalibration shows that the data sets of fluctuation estimates from Ireland, The Netherlands and Britain used in several simulation studies have all considerably overestimated the likely wind fluctuations (Figure 1).
The Irish simulation is based on a similar amount of wind capacity to that yielding the German data. In the space of an hour, the Irish simulation suggests the probability of an 8% change in output over an hour is 11%, while the actual German data reveal the probability of an 8% change in wind output over that short time span is only around 1%. Part of the difference may be due to different wind characteristics, but the data sets from the Netherlands and Britain -- together with real-life operations in Denmark -- confirm the general trend of a far lower probability for significant short term fluctuations than hitherto foreseen.
These smoothing effects of distributed wind are not confined to Europe. Although there are stronger diurnal patterns in some of the American passes, geographical diversity still smoothes the output. This is the case even in California where concentrations of wind turbines are high.
The bottom line
The economic penalties of operating utility networks with increasing amounts of wind arise as additional thermal plant is scheduled to run at part load -- ready to be loaded or unloaded -- to compensate for the rising uncertainty in the supply and demand balance. The amounts of such "spinning reserve" normally scheduled depend on the precise mix of generation. Once the additional reserve needs for wind are ascertained -- by taking into account all the likely fluctuations over a year -- the operational penalties can be quantified from a knowledge of the characteristics of thermal plant.
Studies from a decade ago in Britain and Ireland show a remarkable degree of consistency. Typically, 2% of wind implied a penalty of EUR 0.0014/kWh, and 10% carried a penalty of EUR 0.0026/kWh (Figure 2). The new knowledge from Germany and Denmark, however, is shooting holes in the old estimates and the indications are that the effects of wind variability have been overestimated by a factor of about four. A large reduction in the estimated penalties for wind so far is thus due.
Taking a cautious approach, such as that of a conservative utility, a downward adjustment of at least 33% in the estimation of wind's variability is called for. This will reduce the cost of wind to the system as a whole. That cost is also influenced by the price of coal, since any extra reserve needed on the system to make up for wind's variability will almost invariably be coal plant. Coal prices have dropped recently, but running coal plant on part load is less efficient than full load operation, so that must be taken into account.
On this careful and conservative basis, 10% wind on an integrated system will cost less than one-tenth of a Euro cent per kilowatt hour. This will not be the exact extra cost for all systems, but the methodology used is based on the operating practices of most utilities -- and this is not particularly sensitive to assumptions concerning scheduling accuracy. The exact levels will depend on the particular characteristics of any given system, especially at higher levels of penetration.
As the operating penalties for wind are lower than previously believed, it also becomes clear that any form of dedicated storage to provide back-up for wind is likely to be more expensive than using the existing thermal plant. Storage needs to cover the costs of its operation, the electricity needed to charge it, and repayments of capital. The alternative is for wind to pay the lower costs of keeping existing thermal plant on part load, on the rare occasions it is necessary.
The economics of NETA, however, and other new trading systems, may apparently make it worthwhile to invest in dedicated storage, simply to avoid the unrealistic prices in the balancing market. So storage would be built which is not needed by the system as a whole -- and the consumer would have to be pay its costs.
A better way
The solution to both this problem and the false pricing signals of the proposed balancing market would be for NETA rules to insist that charges be levied only when real additional generation is brought into play -- and not just when "virtual" generation appears on the books as the result of a flawed economic mechanism. Short term accounting for the cost of additional real generation -- more spinning reserve brought into play -- is not possible, but a monthly billing system would be feasible.
Wind, like any other fluctuating power source, or any generator which suddenly has to go off-line, should pay the price of not meeting its forecast. But only when the system operator has incurred costs, and only a fair share of these. As they are, the proposed NETA rules create a market which does not reflect the operational realities for either thermal plant or wind plant.
Wind generators are already looking at ways to get around the problem. Banding together over several geographical locations and offering aggregated wind power could smooth the peaks and troughs of supply patterns from single wind plant. Or wind plant operators could team up with a generator of steady power, just as Northern Alternative Energy has teamed wind and gas to supply Northern States Power (page 18). These efforts, however, are unnecessary when nationwide aggregation exists. They are only being made because wind is having to battle in a distorted market shaped primarily for thermal generators, yet again.