Wind and Sun: Intermittency

Wind and solar power, the two main renewable energy sources, are intermittent and variable. This has two implications: firstly, that the total amount of energy they can produce is less than it would be if they produced continuously at their maximum output; and secondly, that energy storage or some additional source of energy is needed if a continuous supply of energy is required. This doesn’t mean that they’re useless, but it does affect the economics of using them.

The characteristics of the various electricity generating technologies are all different.

Nuclear power has very high capital costs, and low fuel costs. It can run almost all the time, at its maximum output. Because it has high capital costs, it wants to run at maximum output as much as it can, because it costs almost as much when it’s idle as it does when it’s working hard. It likes to supply base load – the power that’s still in demand at the times of minimum demand, usually the middle of the night. At a time when the authorities thought that nuclear power was going to contribute a larger percentage of our electricity than it actually does, great efforts were made to keep base load high: cheaper electricity at night, and encouraging people to use night storage heaters or run their water heating at night; and a pumped storage system was built at Dinorwig, in Snowdonia, to absorb excess power during times of low demand and return it during times of high demand. A pumped storage system is essentially a hydroelectric scheme that can also run in reverse – for which it needs a reservoir at the bottom as well as the top, and pumps as well as turbines (or reversible turbines that can also work as pumps). Dinorwig returns about 75% as much energy from the turbines as it consumes in pumping – not bad for an energy storage system. But the capital cost was high.

Fossil fuel power has much lower capital costs, but high fuel costs – the exact balance between capital cost and fuel cost varies according to the fuel used. For example, gas is generally more expensive than coal (per kWh generated), but the capital cost of the plant is substantially lower. With non-trivial capital costs, fossil fuel power stations also prefer to run at high output as much as possible, but with lower capital costs, they’re much more tolerant of running at low output some of the time than nuclear power. This is particularly true of gas power.

Refuse or agricultural waste combustion power has capital costs similar to fossil fuel power, but lower fuel costs. However, the supply of refuse and agricultural waste is limited, and if the number and size of such power plants increases, then supply and demand considerations would raise the price. (This would of course benefit farmers, local authorities who collect refuse, and other businesses who produce combustible waste.) Again, low capital costs mean that these plants are fairly tolerant of running below maximum output a fair proportion of the time.

Hydroelectric power has medium capital costs, and zero fuel costs – but again, the "fuel" supply is limited. The capital cost of a hydroelectric scheme generally has two main components: the dam, and the generators (some also have significant costs in digging tunnels to take water from the reservoir to the turbines). A scheme can be designed for base load use, with turbines and generators just big enough to use all the water collected in the catchment by running continuously. In general, the extra cost of having much bigger turbines and generators is worthwhile: the scheme can then run at high power when demand is high, and store water for later when demand is low. A base load scheme can be converted later to a peak load scheme by adding more turbines and generators. A peak load scheme has higher capital cost per average output than a base load scheme – but not by a large ratio, and output at times of high demand is more valuable.

Wind power’s capital costs are low in comparison to its peak output, but regardless of demand, wind farms cannot deliver peak output for more than a small percentage of the time. They can deliver some output most of the time, but some of the time they can’t deliver any power at all. Exactly how high the capital costs are in relation to average output depends on the design of the scheme. For a given size of turbine, you can have different sizes of generators: if you fit small generators, the system can deliver peak output for a relatively large percentage of the time, achieving a high availability factor; but in general it’s better to fit larger generators. This results in a reduced availability factor and increases the capital costs, but up to a point the average output increases faster than the capital cost increases, despite the worsening availability factor, because the generators are not the whole of the capital cost. Optimizing this is not simply a matter of maximizing the ratio of average output to capital cost, however: the varying value of the output has to be taken into account. If there’s a large proportion of wind power in the supply system, the value of the output may be lower when it’s very windy.

Solar voltaic panels have a high capital cost and zero fuel cost. The capital cost is coming down, and will almost certainly come down a great deal in the future [Edit, 2017: this has come to pass as I predicted]. The research costs to achieve these reductions are not trivial, but they’re minuscule compared with the cost of the work required to develop new designs of nuclear power stations, and the returns are likely to be much quicker. The issue is again intermittency.

I’ve been writing about capital cost versus fuel costs (ignoring other costs, which are mostly either trivial or have characteristics similar to either capital costs or to fuel costs), glossing over the important issue of fortune telling. Balancing capital against fuel costs is not simply a matter of knowing the future availability factor of a plant, it’s also a matter of predicting (guessing) future interest rates and future prices.

Variable supply of power is basically the same issue as variable demand. If a source can’t supply as much power just now as it usually does, the effect is exactly the same on the system as if a load increased its demand by the same amount. As long as demand doesn’t exceed available supply, you don’t have a problem. If demand is less than available supply, then some suppliers will have to reduce their output. If available supply is less than demand, you can either reduce the supply voltage (a brown out) which makes most equipment consume less power (and correspondingly produce less light, heat, or whatever) but can sometimes damage consumers’ equipment, or you can cut off some consumers (this is called load shedding by the suppliers, or power cuts by the consumers, and is common in poor countries – much less so in rich ones).

Wind and solar power can’t guarantee any output at all, so what’s the point of them?

There are several good methods of Storing Electrical Energy.

Wind and solar power are a good way of reducing the carbon dioxide emissions and fuel costs of fossil fuel power plants, while not being able to replace them completely on their own. Since fuel costs are a large part of the total cost of a fossil fuel station, particularly a gas power station, this is a significant saving.

They are NOT a good mix with nuclear power stations, where very little money is saved by turning down their output. This is of course why the nuclear lobby are so opposed to wind farms and solar voltaic power. (While little money is saved, there is significant benefit in reduced production of nuclear waste – but that’s not something the nuclear lobby like to talk about!)

The same tricks that were devised to compensate for variations in demand for the benefit of nuclear power stations – pumped storage and variable pricing – can also be used to compensate for variations in supply, for the benefit of wind and solar power.

Variable pricing to domestic consumers would ideally be reasonably predictable, so it would only vary with time of day or season in line with expected sunshine and wind, rather than actual sunshine and wind. This would help with the problem, but not eliminate it. Larger consumers, and the utilities themselves, could have variable pricing based on actual sunshine and wind, or more precisely, overall supply and demand. This would help the economics of fossil fuel and waste burning power stations, which would receive a higher price for their power when it was required, compensating for the slack times when they’d get less or nothing. It would also provide the incentive for hydroelectric stations to increase their peak power capability or instal pumped storage facilities.

Even some domestic consumers might be prepared to have less predictable variable prices if it meant lower tariffs part of the time. Anyone charging an electric or hybrid car could benefit for example, and there are some other possibilities as well.

It’s worth noting that variable pricing like this, which favours flexible power supply technologies and renewables rather than nuclear power, is not a distortion of the market, but a liberation of it.

It’s also worth noting that the problem is not as bad as one might at first think.

Individual consumers vary their consumption by large percentages from moment to moment in unpredictable ways, but mostly these changes occur at different times so the average of all of them taken together changes relatively little (in percentage terms) and in a much more predictable way. It would be very difficult for suppliers to cope with the rapid, huge, unpredictable changes of demand that would occur otherwise.

In a very similar way, individual suppliers can change their output by large percentages from moment to moment, as long as they do it independently. It’s not quite as easy to cope with as demand variations, because suppliers typically supply a larger percentage of the total than most consumers consume, and very large industrial consumers can be asked to give warning of their demand variations. But with a good mix of different suppliers, it’s manageable.

Wind power tends to be available at times of high demand for heating; solar power, when it becomes available in quantity, is available during the day, when demand is higher than at night. Wave power, if it is developed, is related to wind power, but the supply peaks will be delayed relative to the associated wind. All of these things help to reduce the size of the gaps that need to be filled by waste combustion, hydroelectric power including pumped storage, other methods of energy storage, and a relatively small remnant of fossil fuel generation.

The amount of storage required in a 100% renewable energy system to cover windless, sunless periods is LESS than the amount that would be required in a 100% nuclear system to even out seasonal variations in demand; windless, sunless periods are much shorter than seasons.

Further reading

For more on the high cost, the low probability of any return, and the long timescales of fusion research see Nuclear fusion? No thanks.

For more detail on availability factor and variation of electricity prices according to supply and demand, see Capacity factor.

For more detail on pumped storage and other methods of energy storage, see Storing Energy.

For more detail on fortune telling – I mean economic forecasting – see Economics again – discounting.