Given the breathtaking speed at which information technology has developed over the past decade, the parallel is an enticing one for renewables. But is it a correct one? The contrast between the two industries is as striking as the parallel: centralised electronic processing is cumbersome, expensive and only available to an elite; centralised electricity generation is efficient, cheap and available to all. Personal computers brought the benefits of main frame computing to the desktop -- good grounds for decentralisation. Small power units, however, bring no tangible benefit to the consumer. That is not to say that electricity generation cannot continue to be efficient, cheap and reliable with a large renewables element, but it raises a question about the necessity of decentralisation.
Radical solutions usually require serious money. A decentralised power system will require investment in central control of small plant; it will need the introduction of clever market incentives and penalties to encourage generation and load management in the right place; it will require advanced planning of the mix of generation; and it will require engineers able to see ten years into the future in order to write the software today that will be needed in a decade.
Perhaps it is not surprising that detailed discussion of the technical and economic viability of decentralised green power systems is noticeable by its absence. Arguments for decentralisation are mostly focused on the failings of large power projects -- economically or technically -- rather than on the reasons why a new system is needed. As a result, power engineers have yet to be seen dashing for their drawing boards in a race to devise engineering solutions for replacement of a well functioning and inexpensive centralised system with an unproved and costly decentralised system to serve the exact same purpose.
The supposition of those advocating decentralisation is that renewable energy can only be fully utilised in a system which is based on local generation meeting local needs. Against a political background in which increased use of renewables and reduced use of fossil fuel is supposedly a global goal, decentralisation would thus appear a must. Yet the nature of renewable energy, which requires electricity to be generated where the source is available, is not necessarily an argument for decentralisation. Indeed, a modified or extended centralised system, rather than the building of a new decentralised system, will allow greater uptake of "distributed" electricity generation, such as that provided by wind energy.
The economic constraint
A real world approach to electricity supply systems of the future requires them to be as stable or more stable than they are today, to have the same or improved security of supply, and to cost about the same, though limited price increases are acceptable in the name of sustainability.
Theoretically, renewables can be integrated in substantial amounts within existing centralised systems without jeopardising reliability or incurring significant additional costs. In practice, such integration depends on the precise mix of plant. A 100% hydro system is quite feasible and not unheard of. Of the so called new renewables, a judicious mix of wind, sun, the waste burning technologies, landfill gas and energy crops (by mass burn or gasification) might also do the job -- if they are each available in the right quantities and if they are fed into a centralised high voltage transmission system.
Whether on a centralised or decentralised system, however, an economic benchmark is going to be the deciding factor for the level of renewables penetration. That benchmark is reached far sooner for a decentralised system than for a centralised one. If money was no object, provision of the necessary quantities of reserve capacity would eventually allow a very high proportion of renewables on any system. But in the real world the public purse is going to run dry long before that point is reached. Even visionaries cannot escape the facts of life.
Wind has high capital costs, as do all the new renewables, and zero fuel costs. Like nuclear, the high proportion of capital cost results in fixed annual charges which recur irrespective of whether the plant is run or not. To avoid sharp rises in the cost of generation, it is essential that the maximum amount of electricity is squeezed from such plant. For this reason, electricity prices from wind plant are much more sensitive to load factors than are gas turbine prices: if the output from wind plant is high when demand is low, and some turbines are being operated only 40% of the time they are capable of running, the price doubles; if Combined Cycle Gas Turbines (CCGT) are put under the same constraint, the price rises by just 30% (fig 1).
No electricity system experiences constant demand throughout the day or in all seasons. The English system's load duration curve, where minimum load is 40% of peak load, is a fairly typical example (fig 2). Indeed, in most centralised systems the ratio between the minimum demand and the maximum lies between 30% and 70%. Taking 50% as a median value, new renewables can thus make up as much as half of the capacity of a centralised system without suffering the cost penalties that incur when lack of demand forces them into part load operation. Fifty per cent is not a limit, because more can be absorbed, but once the capacity exceeds the minimum base load, then the generating costs of the new renewables will rise accordingly.
On a decentralised system the new renewables are far more constrained. The advantages of aggregating power into a central transmission system are lost and the ratio between minimum and peak load is unlikely to be better than 30%. Managing the demand may help, but the higher penetrations achievable on centralised systems are out of reach of decentralised systems without a considerable cost penalty. Realistically, the new renewables can thus make up no more than a third of a decentralised system's power mix. Anything over and above that will have to operate at part load on every occasion that demand bottoms out. With their high capital costs, new renewables then become unrealistically expensive to run. In this respect, advocating a decentralised power supply system for renewables where a centralised system exists is to advocate a poorer market.
Decentralisation does have inherent advantages, however. All power generating plant tends to be sited where it suits the needs of the technology rather than those of the electricity consumer. This is true for coal fired plant, which end up near coal mines, nuclear plant, which head for sites with available cooling water, or wind plant, which migrate to elevated countryside or coastlines. Radical reconfiguring of electricity systems to decentralise them could enable a more geographically balanced -- and hence more stable -- pattern of generation.
Another plus point for decentralisation is that system reliability increases with the number of power plant -- the more generating units feeding into a system the higher the reliability. By its very nature, a decentralised system can only operate with vast numbers of plant distributed at every point of demand, so the "many plant" aspect of system reliability becomes automatically guaranteed. For proponents of decentralisation in the renewables business, much of the lure of the concept is thus closely linked to the idea that small is beautiful: the small unit size of the new renewables makes them ideal for feeding into decentralised systems.
Furthermore, it is claimed that decentralisation will save on transmission losses. Sending electricity across hundreds of miles of grid lines, from outlying power plant to centres of population, or even from one country to another, makes no sense, goes the argument. Local networks and local generation is the answer proposed.
These perceived advantages of radical decentralisation, however, are in fact the actual advantages of distributed generation. Greater geographical balance in plant siting, increased reliability through more but smaller units, and savings in transmission losses are all linked to distributed generation -- the placement of generating plant at more frequent intervals. This can be achieved without decentralisation. Indeed, increases in distributed generation are already happening as part of the natural evolution of the power industry, whether it be with gas units in the United States (see box) or with wind turbines in Denmark.
Distributed generation is desirable on a centralised system -- and for exactly the same reasons as on a decentralised system: stability, reliability and savings in transmission losses. What's more, the beauty of having large numbers of units on an interconnected high voltage system -- as opposed to individual units on a low voltage decentralised system -- is that the loss of one unit causes relatively little upset. Distributed generation also means that fewer spare plant are needed to cope with unexpected outages and uncertainties in load forecasting, which must always be guarded against. To maintain reliability, plant margins must never fall below about 15%. This is a minimum level more achievable if there are many distributed plant.
The essence of running a stable, reliable system at reasonable cost is good control. It is achieved by ensuring that a number of technical criteria are met. The new renewables, in combination and feeding a centralised system with high voltage transmission, can meet many of them. Some criteria, however, are beyond the reach of even the gods of sun and wind -- and on a decentralised system not even the combined forces of all the world's gods could match the required criteria without huge investments.
Reactive power -- required by motors and transformers and a good proportion of consumer demand -- is a major concern to utilities and transmission systems. Normally this is provided by the large steam turbines. These are the units that the proponents of decentralised systems are apparently happy to wave goodbye to. In principle there is no reason why renewable plant should not also fulfil this function, but only hydro and possibly biomass gasification plant are well equipped to provide it at present (table). Wind plant are a particular case. Most use induction generators, which demand reactive power, though variable speed machines can usually supply it. If substantial wind capacity is included on a decentralised system, it will need to be a mix of variable and fixed speed machines.
Frequency control is another vital function which the large steam and hydro plant provide. Like super tankers, their massive inertia keeps them on course, enabling them to resist frequency changes when disturbances occur; alternatively the controls are able to counteract frequency changes. The issue is so vital that the increasing proportion of gas fired plant on the English system has forced the government to declare a moratorium on all new plant. This is because gas is "lighter" than coal on the system and thus has less "head of steam" behind it.
Of the renewables, large hydro is best at providing frequency control -- and the waste burning technologies and biomass plant, again with extra controls, may be able to cope. But there is debate over the ability of wind plant to provide this function. The high inertia of wind turbine rotors is undoubtedly beneficial -- and if the right mix of renewables is available locally, a decentralised system can conceivably provide the level of frequency control needed.
Even so, there is one aspect of control which a decentralised system may find difficult, if not impossible: the ability to "black start" a system without external electricity supplies. After major blackouts, systems need to be re-energised. Generation which can start unaided speeds up this process. Around 25% of plant must have this capability, otherwise systems might never restart. It is uncertain if any of the new renewables can black start a system, with the possible exception of advanced biomass gasification plant. On a decentralised system, plant capable of doing so will have to run permanently on part load, with all the extra cost such back-up entails.
Spinning and smoothing
Mention back-up, and wind and solar inevitably come under fire because of their intermittent nature: they cannot provide power at all the times it is needed. But even the most carefully managed centralised systems cannot exactly match supply and demand, so some plant need to be kept on part load to accommodate demand fluctuations. Provision of this "spinning reserve" is another vital function currently provided by steam plant, but one that will have to be provided by smaller plant on a decentralised system. Although practically all the renewable sources could theoretically provide this function, hydro and biomass are probably best suited to it. But there is the economic constraint of running new renewables on part load to consider; using them to supply spinning reserve will steadily push up their costs to unrealistic levels. On a decentralised system the problem is magnified: the demands of consumers will tend to come all at once, leading to a higher ratio between maximum and minimum demand than on a centralised system. This will require more plant, and even if new renewables could meet the criteria for stability, they would not come close to economic reality.
The larger and more centralised the system, the smaller (in proportion) are the fluctuations due to the mix of consumer types and their geographical spread -- and the more stable the whole becomes. Centralised systems also benefit by pooling their generation. This works to the advantage of wind in particular. When a gale is blowing in North Holland, for example, surplus wind power can be sent south to the more populated regions in the centre of the Netherlands.
The mesh of high voltage transmission links, which in a centralised system carries the power from large plant to demand centres, also serves another function -- it ensures that the reliability of electricity systems is extremely high. If one plant or one circuit fails, alternative paths are available to feed the consumers. All the generating plant is, broadly speaking, available to the whole system. It is not clear how decentralised systems based on distribution networks will be able to maintain reliability without considerable duplication of links -- which incur higher losses than those in high voltage transmission systems -- or the provision of back-up generation. Either way the costs would be enormous, but the gains for the renewables non existent.
Economies of scale
Generating power exclusively where it is needed is what decentralisation is all about. But the savings in transmission losses from not having to wheel power great distances are minimal compared with the extra losses which will be incurred in low voltage distribution systems. These losses rise significantly if renewables are concentrated in areas where local demand is lower than the generation capacity, such as in Wales and northern Germany. On a decentralised system the power is simply lost. With the centralised systems of today a readily available, if costly, solution is to build a high voltage link to send the excess to where it is needed. (A better solution, in terms of system stability, public acceptance and probably cost, is to take a leaf from the distributed generation book and use market mechanisms to encourage a wider geographical spread of generating plant to reduce voltage losses -- but that is another story.)
Even the new renewables will, on occasion, need high voltage links to feed power to centres of demand. If Europe is to meet its 12% target for renewables -- including 40,000 MW of wind by 2010 -- large offshore plant are a likely option. The economies of scale are such that bulk transmission at high voltage may well be needed, especially as not all the best sites are close to demand centres.
There is no getting away from the economies of scale dilemma. Technically it is hard to imagine how decentralised power systems can adequately serve the world's conurbations. Economically it seems totally unrealistic. Office buildings, apartment blocks, industrial complexes, sports grounds and shopping malls can all generate some of their power using distributed local units (PV on buildings seems the obvious choice), but a totally reliable decentralised solution is hard to envisage even if fuel cells became an option. Central London demands 3300 MW at peak periods (incidentally, this is when the sun is asleep), most of which is wheeled in from miles away. Of this demand, a combined cycle gas turbine station on the outskirts now provides 1000 MW.
On the subject of economies of scale, CCGT costs provide a salutary lesson that big might not be beautiful, but it is cheap. CCGT costs decline rapidly with increasing size (fig 3) because of higher efficiency and lower capital costs. A 400 MW CCGT unit is around 55% efficient and in the UK can produce electricity for around $0.032/kWh, a target which is hard to beat. The corresponding generation cost for a 5 MW gas turbine is almost double. So, despite the fact that electricity from centralised stations often needs to be transported -- possibly some distance -- to the consumers, the associated transmission losses and costs pale into insignificance beside the extra expense of decentralised gas plant.
The intermittency myth
A commonly held, but mistaken belief, is that the inclusion of intermittent renewable energy plant on a system will require the construction of extra stand-by plant because wind and solar cannot generate at all times. This is true, to a small degree, on a centralised system with a high penetration of renewables plant. But if it is true for a centralised system, logic dictates that the problem would be far worse on a decentralised system.
Intermittent renewable energy technologies feeding into large interconnected power systems have a major advantage over those feeding into a decentralised system: fossil fuel plant already provide back-up capacity, or "spinning reserve," to accommodate rapid changes in demand or errors in load prediction.
Back-up is needed everywhere for every kind of plant. Sources such as wind only necessitate changes in operating strategies if they raise the overall level of uncertainty in the generation/demand balance. This is unlikely to be the case with wind for a long time to come. Even with the relatively high amounts of wind in parts of Denmark, no specific reserve capacity has been built. The rising wind capacity has not sparked a need for more back-up because more wind inevitably means greater geographic dispersion, which smooths out the fluctuations; in any country it may be windy in one region and not in the other, so sharp fluctuations in wind plant output -- of the order of hundreds of megawatts or more -- are unlikely to occur.
If wind reaches levels higher than 20% of maximum demand, additional thermal plant may need to be kept on part load to cope with fluctuations in output. The bottom line -- the extra cost of running the thermal plant (fig 4) -- is very modest at about $0.001/kWh with 2% of wind on the system, rising to around $0.004/kWh with 15% wind energy (roughly equivalent to 30% of capacity). This is a tiny amount compared to the provision of purpose built back-up or "dedicated storage," an option frequently canvassed and the only solution available for a decentralised system wanting to retain reliability.
The precise levels at which wind becomes difficult to absorb on a system depends on the characteristics of other plant, particularly the amount of hydro or pumped storage (which can respond to changes in wind output) and of nuclear and combined heat and power (both of which tend to be rather inflexible). What is clear is that the vagaries of the wind or the sun are not a problem for centralised systems. Greater threats to stability abound, particularly the loss of interconnectors or of large thermal plant. Intermittency, however, will be a problem for decentralised systems. With less (or no) transmission net, they cannot call for power being generated tens or hundreds of miles away, where the wind is blowing or the sun is shining.
For the sake of incorporating the most wind power as possible on a system, and for the sake of system stability, the good load-following capabilities of fossil plant provide the best basis for integrating large capacities of wind or sun.
Evolution not revolution
Adding renewables to the centralised systems of the western world will often prove cost effective compared with adding new central plant. There is no reason, economic or technical, why development of the renewables should not proceed in parallel with the continuing operation of high efficiency power systems based on fossil plant.
Seen from the point of view of the wind industry, gradual introduction of distributed generation -- a utility evolution now in progress -- is preferable to any revolutionary restructuring to decentralise power systems. Even with 50% of new renewables no unrealistic increases in cost are incurred. Such a system will enable large fossil plant to feed large towns and meet basic needs, such as spinning reserve and frequency control, at economic cost.
On a decentralised system, if it is to be stable, renewables will either be more expensive because they need their own back-up, or limited because there is no money to spend on back-up. Pursuing decentralisation for its own sake has little merit, especially as the potential of larger new renewable installations, such as offshore wind and tidal power, will go to waste without high voltage transmission lines to take the output to where it is needed.
As renewables increase their output, however, changes will be needed to today's systems. In countries like Denmark and Germany, where penetrations of wind in some regions are already around 30%, utilities are modifying systems to cope with the new forms of generation. Small plant will not be able to avoid becoming subject to central control at some point. Unless controllers know how much to expect, they cannot run an efficient power system. The demand/generation balance can usually be predicted with an average error around 1.5%. If a system has, say, 10% of renewables, but the controllers are unaware that, say, 1% is out of action due to faults or maintenance, then the prediction error goes up to 2.5%. So "subject to central control" does not imply that an eagerly turning wind plant is liable to be switched off by utility controller with too much power on his hands. It simply means "ready to notify of availability."
Tighter regional control may be appropriate, with the central controls instructing regional controllers on parameters needed to keep the whole system stable. Whichever route is adopted, investment in appropriate technology is likely to be needed if small renewable plant is to be controlled from remote locations. Manned controls at each renewable plant would be excessively costly.
One vital prerequisite for stable and efficient power systems is the need for sensible siting -- from the point of view of the electrical systems. Broadly speaking, power plants are sited so as to deliver electricity at minimum cost (and maximum profit) to the generator -- and the renewables are no exception. As concentrations of renewables rise in areas of low demand, losses in low voltage systems increase and costs rise correspondingly. A balance needs to be struck between the operation of efficient power systems and the needs of generators.
"Locational price signals" are one answer, though in networks where they have been tried, such as in England and Wales, they have not been conspicuously successful. Transmission systems, for all the reasons stated here, cannot just be adapted to suit renewables -- not without losses of stability and reliability, not to mention unacceptable (and unnecessary) increases in cost. Strategic siting is a better and cheaper option. A further, non-technical and non-economic, advantage of such a policy for wind is that easing development pressure on the windiest regions will also alleviate public pressure against the development.
Into the future
Renewable penetration at even higher levels, those above 50%, are quite possible, but they incur extra costs. These can be forecast -- and contained -- but the steadying influence of the large coal plant, or hydro where this is an option, is vital. It is an important point to remember before continuing to lobby for decentralisation -- and thus the abolition of the very plant which will allow renewables to meet their full potential. Energy crops might substitute for coal plant in this respect -- for control and part-load duties -- but biomass gasification plant is in its infancy and nobody yet knows if it can fill such an important role.
Once over the 50% penetration mark, it is not realistic to expect that all renewables can enjoy their favoured "must run" status indefinitely. Technical and economic constraints will apply and part-load operation must be accepted in due course. Evolution will see to this as distributed generation continues on its current course. A revolution in the name of decentralisation will only hamper progress.