It is often followed by the argument that electricity networks using wind will struggle, their stability will be compromised and they will incur huge extra costs.
This is wrong. Modest extra costs are incurred, but on nothing like the scale suggested by some.
Another common misconception is that you can divorce the issues surrounding integration of variable renewables such as wind from the "normal" workings of an electricity network. You can't — integration is exactly what is required.
Any power system — the network across which all power is distributed — must deal with national and regional fluctuations in power demand, and balance that with the power available, including fluctuating wind. To find out what effect wind has on the operation of an electricity system, we can match consumer demand with wind power variations. In the past, discussions about assimilating, say, 20% wind into an electricity network have been quite theoretical, but this now reflects reality. This level is being achieved or anticipated in many areas, as the map on page 16 shows.
As consumers go about their day-to-day business they make frequent changes in their demands for power. Pooling the demands together in an electricity network helps to smooth out fluctuations. But, while domestic consumers in every European country do not all switch on their electric ovens at the same time, there are still substantial demand fluctuations on power networks.
During a typical day, the "baseload", or minimum level of power demand, occurs in the early hours of the morning. From about 5am until 10am, demand increases from domestic and commercial consumers, as well as from industry. In winter in the UK, for example, demand increases from around 35GW to 44GW during these hours. In western Denmark, a much smaller system, it increases by about 1.3GW (see graph, below).
In sunny weather demand may continue to increase, driven by air-conditioning systems, and peak between 12pm and 3pm. In most of northern Europe, however, the daily peak in winter tends to come at around 5:30pm, when transport demands for the rush hour coincide with domestic heating and cooking demands. Maximum demand in the UK is around 60GW.
In addition to these largely predictable demands for power, unpredictable changes occur, some quite short term. These may be caused by cloud or heavy rain, or even by television programmes that turn out to be more or less popular than the system operator had predicted. These operators continually issue instructions to the power stations to increase or decrease their output. But power stations can break down, sometimes through mechanical fault, but more frequently through instrumentation fault, meaning several hundred megawatts of power generation can vanish from the network instantaneously. That really is "intermittency", a drawback often wrongly ascribed to wind.
System operators have vast databanks of information on likely demand variations yet cannot predict them with 100% accuracy. They certainly cannot predict when steam turbines or nuclear reactors are going to break down. To cope with mismatches between supply and demand, a system will always have reserves from a supply that can increase or reduce output accordingly. Reserves are typically set at around 5% of maximum demand (so the UK will have 3GW of plant in reserve). Reserves can come from any type of generating plant whose output can be adjusted either automatically or on request by the system operator. They must operate at less than full output in order to be able to increase output when requested; with most thermal plant this means reduced efficiency, so the plant owners need to be recompensed for the higher fuel costs.
Just as demand fluctuations can be evened out across a larger network, so can wind-power production. One wind farm can smooth the total output of its turbines, but the total output of all the wind turbines in Germany, for example, is much less variable than the output from a single site. Interestingly, comparisons of national wind-power performance data swings show similarities across European countries. The maximum swing in output over an hour rarely exceeds 20% of the rated capacity of the wind plant. So wind, contrary to the popular notion, is not totally unpredictable — and it is variable.
An enormous amount of work is in progress to predict wind-power production, but system operators can already analyse fluctuations in order to manage their system. For example, if the output from the 6GW of wind now installed in the UK is 2GW at noon, then the output one hour later is likely to be between 1.82GW and 2.18GW. It is extremely unlikely that it will be less than 800MW or more than 3.2GW.
The cost implication
Calculating the extra uncertainty of adding wind to a system is not a simple case of addition. For example, if the uncertainty in the demand forecast one hour ahead is 400MW and the uncertainty in wind output for the same period is 150MW, the combined uncertainty is not simply 550MW (400+150) — a more complex statistical calculation returns the lower uncertainty of 427MW.
Armed with this information, system operators are in a position to know how much extra reserve to schedule when their system includes wind. To cope with that 27MW additional uncertainly, they would probably schedule three times more — 81MW. The cost of that can then be quantified, usually using market rates for extra reserves. The level of these extra costs varies between utilities but is invariably small — in July 2011, Windpower Monthly quoted between $1 and $7/MWh (€0.7-5.2/MWh) of wind generation for 20% wind.
David Milborrow is technical consultant for Windpower Monthly