One of the side effects of electricity liberalisation is that the intermittent renewable sources often suffer. Take electricity generation and supply out of the hands of state control and it invariably means greater fragmentation of the industry -- to promote competition. But fragmentation undermines one of the cornerstones of an integrated electricity system -- the benefits of aggregation. The greater the aggregation, the greater the smoothing effect on peaks and troughs in supply and demand, and the better the system can absorb variations in supply from intermittent sources of electricity like wind power.
A striking illustration of the more unpleasant side effects of the liberalisation process is the dire impact on wind power of the UK's New Electricity Trading Arrangements (NETA). Such are the effective penalties for not supplying power to a strict schedule that the best way to operate a wind farm under NETA in the UK is to shut it down (Windpower Monthly, May 2001). News of this kind reinforces the widespread, but mistaken, belief that operation of an electricity system with wind energy -- or the other intermittent renewables -- causes problems. It need not.
There is a double irony in the news from the UK. First, the grid operator expects to schedule more reserve plant to secure continuous supply under the new trading arrangements. This means that coping with any trivial perturbations from the existing wind plant -- which never was going to be a problem -- becomes even more irrelevant. Second, wind's entrapment in a vice of bureaucratic making in the UK comes at a time when data from Germany and Denmark show that fluctuations in output from wind plant are significantly lower than most simulation studies have predicted.
At the same time, Denmark is forging ahead with plans that will have 50% of its electricity coming from wind energy by 2030. If nothing else, this alone seems to be a strong indication that the "problem" of a fluctuating power source is not quite as bad as is often made out. Last year Denmark got around 17% of its power from the wind
What happens when the wind stops blowing? On a small island, fed solely by wind, the lights will indeed go out unless there is some back-up. In an integrated electricity system, however, there is plenty of back-up already in place -- not provided solely for wind.
Integrated electricity systems are not wholly dependent on one source of supply. Moreover, there are numerous other threats to stability which are far more serious than wind fluctuations. The relevant issues are aggregation, operational issues and system security.
Before dealing with these issues, an important aside: there is a vital distinction which must be drawn between the technical issues and accounting convenience (read NETA or the rules under which electricity is traded in Denmark). Overall costs of electricity systems are minimised in a technically optimised system. Ideally, the rules of an electricity market should deliver this. In practice, however, administrative rules get in the way of efficiency. One of the unfortunate spin-off effects is that wind energy is unfairly penalised.
Aggregation v home supplies
Integrated electricity systems make good sense. They minimise the need for power plant and are far more efficient than the fragmented systems they replaced. An aim of liberalisation should be to retain high levels of efficiency, if not for economic reasons then for the sake of the environment. The UK and its 20 million households has gone furthest down the liberalisation path, but if each of those households had its own power supply it would need to be sized to cope with the maximum demand from each dwelling, 5 kW, say. So 100 GW of power would be needed to cope with domestic demands alone, whereas total electrical capacity in the UK is just over half that. The greater the aggregation, the smaller (in proportion) are the variations in demand and the easier it is to predict them.
At one end of the spectrum the minimum demand from a domestic dwelling is a few watts, the average is about half a kilowatt and the maximum is in the 5-10 kW range -- ten to 20 times the average. Aggregation smoothes these variations in demand from all sectors -- domestic, commercial and industrial. Nationally, the maximum demand is 58 GW, about 1.5 times the year-round average demand of about 40 GW. As demands are added and smoothed, prediction becomes easier; the National Grid Company (before NETA) could predict the match between supply and demand at each half-hour with an average error of about 1%.
Integrated electricity systems smooth the many variations in demand and generation because the individual fluctuations do not all occur at the same time. When added together, ten randomly picked consumer "demands" on a system reveal a range in fluctuation of just 40-70 kW. Yet the individual range leaps around between 0 kW and 10 kW (figure left). The random numbers could also represent output from each of ten wind farms, again illustrating the point that aggregation makes systems less susceptible to fluctuations and easier to manage.
From an operator's view
System operators cannot detect variations in output from modest amounts of wind plant since they are swamped by numerous other fluctuations of similar or greater magnitude. The output from a 5 MW wind farm might be curtailed in high winds over a period of a few minutes, but a Eurostar train which shuts off power and coasts "switches off" a 5 MW load in a matter of seconds.
As wind capacity increases, however, it will eventually increase the uncertainty in the supply/demand balance. Critics of wind energy say more "spinning reserve" (part-loaded thermal generation with output that can be increased or decreased) needs to be scheduled to cover for fluctuations in output from wind plant. The argument is erroneous until there is significant wind penetration on a system. The level of wind capacity at which the uncertainty in the supply/demand balance is increased is the key issue. Practical experience with large amounts of wind on a system is almost non existent, so theories have dominated the debate. Western Denmark, however, with about 1750 MW of wind, and Germany, where the output of 350 MW of wind is monitored continuously, are providing some answers (box).
So far the indications are all positive: large amounts of wind can be absorbed in existing systems without upsetting the balance. In other words, real world practice is proving that fluctuations in wind power output are considerably less than those predicted by simulation studies in The Netherlands, Ireland and the UK.
Conditions for wind power integration set by system operators can either level or tilt the playing field. The UK's National Grid Company (NGC) recently specified general criteria for the generation it will accept -- and these criteria encouragingly allow for a deal of power from an intermittent source like wind. NGC says fluctuating generation can make up 20% of peak demand, that the potential for instantaneous loss of generation can be up to 2% of peak demand, and that the potential loss of a generation source in an hour can be up to 3% of peak demand.
Armed with NGC's bottom-line for intermittent generation -- and the knowledge gleaned from day-to-day operation of large amounts of wind power in Germany and Denmark that wind fluctuations are less than predicted -- the industry is in a position to dispute the legitimacy of less well-founded criteria. The fact that the largest power excursions within an hour are unlikely to exceed 20% of wind capacity is a powerful argument.
Given that the maximum demand met by NGC is around 50,000 MW, the first criterion suggests that penalties will be incurred when the capacity of wind exceeds about 10,000 MW. Moreover, the "allowable" power excursion within one hour is set at about 1500 MW on a 50,000 MW system. Since the maximum hourly power swing from well-distributed wind energy is proving to be around 20% from the German and Danish experience, irrespective of installed capacity, this suggests that 7500 MW (NGC's figure of 1500 MW, multiplied by five) of wind could be installed in the UK before extra reserve -- with its associated costs -- needs to be scheduled. This is a conservative figure and not a ceiling, but a rough marker to indicate where operational penalties due to wind fluctuations may start to become significant.
The commercial aspect
The 7500 MW of wind is 15% of NGC's peak demand, whereas the wind plant installed in western Denmark make up roughly 40% of peak demand. A look at what is happening there should give a clue to the likely market-based penalties wind can expect. But the system operator seems to exploit the links with Norway, Sweden and Germany to export most of the wind generation (see figure above) and so the question of system penalties for additional spinning reserve does not arise. The system of pricing under which this power is traded are complex, but the costs to the system operator were around DKK 65 million in 1999, which corresponds to EUR 0.004/kWh of wind energy. Although this may be more than the technical cost of operating additional thermal plant it reflects no more than the market rules, including the complex and arbitrary payments for renewable energy in Denmark. Significantly, the wind power is being traded with systems having substantial hydro plant and so the net effect is likely to be reduced greenhouse gas emissions.
The level of spinning reserve on any system depends on a number of factors, such as the load flows and the levels of water in pumped storage facilities. The combination of available pumped storage, plus spinning reserve, must be sufficient to keep the system stable and frequency within the target range in the event of a sudden loss of generation or a main transmission circuit failure. In England it is generally in the region of 200-600 MW and in eastern Denmark -- although a much smaller system -- it is similar. The reason for this is that the Danish system needs to guard against loss of its biggest unit, which has a capacity of 600 MW.
With large or small systems, therefore, the normal reserves are more than adequate to cope with wind fluctuations -- until the day that wind becomes a significant supply source. When large power stations trip out, the benefits of aggregation again come into play, as the enormous mechanical and electrical inertia in the system means that there may be a slight lowering of frequency and voltage, but rarely sufficient to be detectable.
Wind energy is unlikely ever to be a threat to system security. It is very difficult to visualise a scenario whereby 1000 MW of wind output disappears instantaneously. It is quite inconceivable that a sudden gust of wind could engulf the whole of Denmark, say, so that every single machine reached its cut-out speed and shut down simultaneously.
Although few technical problems stand in the way of increasing wind energy penetration into electricity networks, the optimistic plans for expansion in both America and Europe must take two factors into account. First, although integrated electricity systems can easily handle substantial quantities of wind -- at modest cost -- the location of that wind is all-important.
Geographical diversity, which smoothes the output, is essential but, equally important, a careful watch must be kept on the location of the associated thermal or hydro plant likely to be needed as additional reserve. One of the issues being studied by Nordel, the Scandinavian grid operator, is possible constraints imposed by the capacity of the links between Denmark and its neighbours. The same study confirms, however, that the need for additional reserve, even if wind accounts for 25% of maximum demand, are modest.
The second problem is a tricky one. Politicians and energy regulators need to be convinced that wind energy -- and the other intermittent renewables -- pose no threat to the operation of their electricity systems. In other words, there is no need to saddle them with unrealistic operational penalties, as is happening today. The fundamental problem is that markets do not properly acknowledge the benefits of aggregation and the penalties likely to be incurred by wind are not due to any real technical problem.