Environmental Audit CommitteeWritten evidence submitted by Malcolm Grimston

THE FULL COSTS OF GENERATING ELECTRICITY

Executive Summary

The recent debate about costs of low-carbon electricity has tended to be based on a rather crude “cost of a generating unit divided by number of units produced” approach. The potentially vast costs associated with managing the inherent intermittency of some renewables, notably wind and solar, are not at present fully reflected either in the stated costs of these energy sources nor in the quoted “subsidies” they are receiving. It is becoming increasingly clear that the full costs—both economic and environmental—of all sources of electricity, most notably variable sources such as renewables like wind and solar, should include not only the direct costs referred to above but also indirect costs caused by, for example, the intermittency of the output. These costs are potentially very large and include the need for grid strengthening (“grid-level system costs”); the effect on the economics of other generators, leading to higher prices (or threats to security of supply, in itself a huge potential cost to consumers); and the greenhouse gas emissions caused by the need vary the output of dispatchable fossil-fuelled power plants, thereby reducing their thermal efficiency. A rational approach to allocating limited resources to deliver maximum supply security and emission reduction would involve ascribing all costs and environmental effects, direct and indirect, to the source which has ultimately given rise to them. This would lead to a very different discussion re the way of achieving economic, security of supply and environmental goals—for example, arguably leading to a reclassification of variable renewables as “medium carbon”.

The Unique Challenges of Electricity

Electricity is a unique commodity. Despite over a century of research, it still cannot be stored in significant amounts—pumped storage is feasible in some circumstances but requires appropriate geology and major capital investment and suffers from losses of some 15%–30% of the power input.1

For most commodities, of course, it is possible to manufacture them when it is convenient, then to stockpile them until they are needed. If it makes sense, for example, to manufacture shoes overnight (say to take advantage of low power prices), the shoes can be stored for a practically unlimited time until someone wants to buy them. But with electricity we have to be able to make exactly the right amount to fill requirements on a moment-by-moment basis. Generating too much puts great strain on the wires which carry electricity from the power station to the customer and can lead to their deforming and ultimately melting, with serious effects—eg “electrical arcing” or severe damage to transmission capacity. Generating too little puts us in danger of poor quality electricity supplies and, eventually, power cuts, which are extremely expensive. Among the effects of power cuts are failures in our transport systems, losses of a day’s production in the workplace, loss of a freezer full of food, inability to pump water into our homes or power our hospitals, and severe social disruption, eg looting and vandalism if the power cut occurs overnight.

Even if the discrepancy between supply and demand is not sufficient to cause damage to the transmission network or power outages, an excess or shortage of power demand can result in an increase or decrease in frequency of the power supply. (In the UK the National Grid has a statutory duty to maintain the frequency in the range of 49.5 Hz to 50.5 Hz and usually upholds it at between 49.8 to 50.2 Hz.2) If the frequency should become too high this can cause direct damage to a variety of appliances and/or increase wear and tear. If the frequency is too low then equipment will tend to underperform. In turn this may lead operators to increase their demand where they can (eg the driver of an electric train opening the throttle to compensate for the lower speed), which may exacerbate the supply/demand discrepancy.

VARIATIONS IN UK POWER DEMAND, 2010–113

Power demand varies considerably, both during the day and during the year. The difference between peak demand and lowest demand in the UK is roughly a factor of 3 (60,000 MW to 20,000 MW). Usually demand fluctuation is fairly predictable but variations can be very rapid. The biggest power surge to date in the UK was some 2,800 MW after the penalty shoot-out in the 1990 football World Cup semi-final between England and West Germany in 1990 (when supporters put the kettle on for a consolatory cup of tea). Electricity planners needed to be ready for the moment when coverage of the Royal Wedding in 2011 reverted to the TV studio, when demand surged by 2,400 MW.4

This balancing act creates considerable challenges, both technical and economic. In a functioning system, the grid operator, who is ultimately responsible for keeping supply and demand in balance, has to be able to order (or “dispatch”) enough reliable electricity to fulfil demand, but no more. Furthermore, somebody has to keep power stations available to fill very high levels of demand, say on a cold, still, cloudless winter evening, while knowing that these plants will not be called upon for most of the rest of the year when demand is lower. The costs of failing to fulfil demand are very high—typically the “Value of Lost Load” (VOLL) is estimated at between 50 and 350 times the price of a delivered unit.5

The Importance of “Dispatchability”

Traditionally, most countries have overcome these problems by building, or encouraging the building of, large power stations whose power can be varied (or which can be turned on and off) in response to changes in demand. Nuclear power stations, which have very low variable costs—in other words, since they use very little fuel they cost very little more when they are operating compared to when they are not—tend to be used for “baseload”, ie the power demand that never goes away (to keep water pumping systems operating, transport running and so on). The baseload is roughly 20 GW in the UK, considerably more than installed nuclear capacity of around 10 GW, meaning that when the system is being run rationally the nuclear stations will operate continuously whenever they are available. Fossil-fuelled power stations—particularly gas but also coal—are used to “load-follow”, since a considerable proportion of their costs is saved when they are not running. However, the important thing is that all of this plant is “dispatchable”—ie it will be available to generate if the grid operator requires it to do so (unless it is closed down for maintenance or has broken down unexpectedly). At times of very high demand the real-time power price increases enormously to compensate those companies which do keep power stations ticking over to fulfil that very high demand.

In order to maintain the quality of supply and ultimately to keep the lights on in unforeseen circumstances—say a particularly cold snap or a time when two or three large power stations break down unexpectedly—the grid operator seeks to maintain what is called a “capacity margin” in the system as a whole. So if we expect the highest demand in a particular year to be about 60,000 MW, the ideal situation would be to keep about 72,000 MW of power capacity ready for use, in case the demand should happen to be say 65,000 MW and up to 7,000 MW of power plant was undergoing maintenance or had broken down unexpectedly. A capacity margin of about 20% is generally regarded as sufficient, though from time to time it has exceeded this (and indeed fallen below it).

The Challenge of Variability

However, recently things have begun to change significantly because of growing use of renewable sources of energy, many of which are “variable” (the opposite of “dispatchable”) because they depend on unpredictable, or at least very variable, weather conditions. The output from, say, a windfarm will typically be about 25–30% of its “rated” capacity—in other words if the windfarm produces 100 MW of electricity when the wind is blowing at the best speed, the average output will be between 25 and 30 MW. (In 2010 the figure was 23.7% in the UK, in 2011 it was 29.8%.6) However, it is very difficult to predict with any great accuracy or much in advance when the wind will blow—indeed, wind speeds can vary over a few seconds or minutes on a “gusty” day.

SHORT-TERM VARIABILITY IN WIND SPEED7

The output can vary very rapidly—sometimes within a matter of seconds or minutes—as weather conditions change. Although the most rapid variations are to some extent compensated for by the inertia of the wind turbine rotor, the phenomenon still creates technical, financial and economic problems. Technically, turbulence necessitates the provision of a great deal more wire capacity in order to accommodate all power being produced when the wind is blowing, but then very rapidly switch supply to another area of the country when the wind drops and other types of power plant are required. In China about half of the energy being produced in windfarms is wasted because the grid connections are not yet in place.8 It also offers challenges to maintain system frequency and imposes more wear and tear on the wind turbines. Wind turbine towers are usually tall enough to avoid the greater wind turbulence encountered close to ground or sea level.

Evidence suggests that low wind speed tends to be weakly correlated with high power demand (cold windless winter evenings and hot windless summer days), further exacerbating the challenges.9 At other times wind or solar generators may be producing close to 100% of their rated output, risking overloading the system and obviating other power sources to close down.

VARIATION IN OUTPUT FROM WIND GENERATORS IN GERMANY, 200310

It is not enough simply to have sufficient generating capacity available to meet demand. If the quality of the supply (especially its frequency) is to be maintained then the capacity available must be able to respond quickly to changes in demand on the system. With the exception of some renewables such as hydropower and biofuels, which are dispatchable, most renewables are unable to provide this responsiveness as the output is determined by wind speeds, tides or sunshine. The output of conventional power stations, by contrast, can be actively varied by increasing or decreasing the amount of steam being generated and fed to the turbines.

Managing a system which includes a lot of variable renewables presents a number of challenges, depending on whether the wind is blowing or is not.

When the wind is blowing (or the sun is out or the tide is in), in order to prevent excess supply and, for example, damage to the transmission wires or unacceptably high frequencies, quite a lot of that energy has to be wasted—in the UK in recent years wind generators have often been paid quite large sums of money to stop generating electricity when the wind is good11—and/or other companies have to cut their output to compensate.

The latter response undermines the profitability of the gas- or coal-fired plants that are having the market cut from under them. (The economics of nuclear power means that it would be the last to be withdrawn if there was an excess of variable output at any particular time.) But when the wind is not blowing we still need almost as many gas- and coal-fired power stations as we would have needed if we have never built the windfarms in the first place. So some way has to be found of compensating the fossil fuel power stations for the market they lose when the wind is blowing. If this is not done they might be closed down or mothballed, and so not be available for when the wind drops. This problem is already being seen in countries like Germany and Spain with a lot of wind energy12 and is also at the heart of some of the contradictions in the UK Energy Bill which was published in draft form in December 2012.

On the one hand, the Energy Bill is responding to the likelihood that, left to a free market, investors would only fund new build combined cycle gas turbines (CCGT). CCGT is quick and cheap to build and maintain and the technology is particularly appropriate for load-following, it being relatively easy and economically worthwhile to vary its output depending on the system demand. Although it is expensive to run (fuel costs dominating overall costs) it is economically quite low risk, because if the gas price does soar (the key economic sensitivity for gas-generated electricity, since most of the cost of that electricity is the cost of gas), then the extra cost can very largely simply be passed on in higher power prices to customers who have nowhere else to go, at least in the short term.

Nuclear and renewables, by contrast, are much more capital intensive (ie they cost a lot more to build per unit of installed capacity), though they are cheaper to run. Unmitigated coal capacity lies between CCGT and low-carbon sources, while coal with Carbon Capture and Storage (CCS) also has high capital costs. The key economic sensitivity for these sources, then, is how the initial construction programme is managed. The costs associated with any time or cost overruns in the construction phase cannot be passed on to consumers in the same way. Consumers could simply commission a new CCGT to replace the output from the nuclear or renewable plant while it was being finished. So the Energy Bill proposes a “carbon price floor” (ie a guaranteed bonus for sources of electricity like nuclear and renewables that do not emit carbon dioxide) and contracts-for-difference (in effect a guaranteed long-term power price) to promote investment in renewables and nuclear.

On the other hand, however, if these measures work and more variable renewables are brought online, then as noted earlier the economic case for investing in new CCGT or other dispatchable technology is weakened, since these plants would be left without an income during those times when the wind was blowing at the right speed. (This is even more the case with more heavily capital intensive sources such as coal with CCS or nuclear, as their economics are more seriously harmed by being taken off line.13) But eventually we have to have new CCGTs (or other flexible dispatchable capacity) to provide power when the wind drops, or the lights go out. So the government is having to introduce other measures, notably capacity payments (paying companies to keep power capacity available even if it is not being used), to compensate CCGT for the effects of the measures it introduced to deter CCGT in the first place. The vast costs incurred by this need to manage the inherent intermittency of renewables are not at present taken into account when calculating the true costs of renewables and the full subsidies they receive. German energy and environment minister Peter Altmaier has estimated the cost of Germany’s transition to renewables at up to €1 trillion. Feed-in tariffs account for more than half of the total, underpinned by a regime under which customers are forced to take all the renewable electricity being generated at any particular moment. This “must-take” regime represents a major subsidy to renewables alongside the upfront subsidies but is rarely described as such. Improvements to the German grid will cost an estimated €27.5 billion and €42.5 billion.14 Over 8,000 km of new or upgraded transmission lines (with associated environmental impact) will be required and grid operators have said, “The investments required for expanding the transmission network only represent a fraction of the energy switchover’s cost, but they are essential for its successful implementation”.15

The point is emphasised by a table from the UK government’s National Policy Statement on Energy (July 2011).16 Already the UK is running a significant capacity margin (over 40%) because an increasing amount of that capacity is variable and cannot be relied upon to fulfil peak demand. The projected need for generating capacity in 2025 is no less than 113 GW, to cover peak demand unlikely to be more than about 65 GW in that year, even with good economic recovery. That would seem to represent a capacity margin of over 70%, vast by historical standards. It would be necessary purely to cope with the variability of the 33+ GW of new renewable capacity. Whether the cost of building and maintaining such enormous redundancy is borne by the renewables or, by distorting the market, is apportioned to other players (through “must take” contracts for renewable output), it is ultimately going to land on the backs of consumers (and/or the taxpayer).

Total current generating capacity	85 GW
Peak electricity demand now & 2020	60 GW
Average demand	43 GW
Closure of coal plants by 2015 owing to the Large Combustion Plant Directive	12 GW
Nuclear closures over next 20 years	10 GW
Generating capacity required in 2025	113 GW
Of which new build	59 GW
Of which renewable	33 GW
For industry to determine	26 GW
Non-nuclear already under construction	8 GW
Current proposals for new reactors	16 GW

So when it comes to looking at the costs of various sources of electricity, it is highly misleading simply to consider the cost of installing the wind generator or solar panel divided by the number of units of electricity it produces. Some sources of electricity, notably the variable renewables, impose huge costs on the system as a whole.

Grid-Level System Costs

The issue is considered in depth in an OECD/NEA study published in November 2012.17 The report looks at the way various methods of generating electricity interact with each other. Such external costs of particular sources, or “system costs”, are defined as the total costs above plant level costs to supply electricity at a given load and given level of security of supply. They can take the form of intermittency, network congestion, greater instability (ie higher risk of interruption to supply) etc. The focus of the report is on system costs associated with nuclear power and renewables such as wind and solar photovoltaics. In particular, the report considers “grid-level system costs”, a subset of overall system costs that consists of the costs of network connection, extension and reinforcement, short-term balancing and long-term adequacy in order to ensure continuous matching of supply and demand under all circumstances.

The report considers the complex issues arising from the integration of significant amounts of variable renewables, which profoundly affect the structure, financing and operation of electricity systems, thereby having economic and financial implications well above the “plant level” costs of these sources. It includes the first quantitative study of grid-level system costs in six countries (Finland, France, Germany, South Korea, UK and USA).

The most important effects of a large variable renewable component in a particular electricity system may be:

lower and more volatile electricity prices in wholesale markets due to the influx of variable renewables with very low marginal costs (including zero fuel costs), leading to the closure or mothballing of existing plant need to maintain secure supplies;

the reduction of load factors of dispatchable power generators (the “compression effect”) as renewables with very marginal cost are given priority over dispatchable supply;

the introduction of inefficiencies in existing plants coupled with an influx of renewables, implying an increasing gap between the costs of producing electricity and prices on electricity wholesale markets; and

greater physical wear and tear to thermal power plants owing to the greater stress on (especially) metal components—as the metal expands and contracts alongside unnecessary increases and decreases in output and hence core operating temperature it is more likely to crack—thereby shortening the life and/or increasing the maintenance costs of these plants.

The financial implications of variable renewables are therefore potentially profound in both the short term and the long term.

In the short term, the dispatchable power technologies will suffer owing to the compression effect. The effect on nuclear may be less than on other dispatchable technologies, as nuclear has low variable costs. It is therefore likely to continue to run when large amounts of renewable electricity are available (eg because the wind is blowing at or near optimum speeds) as there is little economic advantage in taking them offline (while taking CCGT offline saves significant amounts of gas). As noted earlier, gas plants are already experiencing substantial declines in profitability in many countries with high shares of variable renewables.

The threat to security of supply represented by the encroachment of renewables into the market of dispatchable generators are amply demonstrated by recent developments in the UK. Some 4.5 GW of coal-fired capacity at Didcot, Kingsnorth and Cockenzie, and over 2 GW of oil-fired capacity at Fawley and Grain, was closed earlier than expected in 2012–13.18 At the same time capacity margins are projected to fall dramatically in the near future, perhaps as low as 4% by 2015–16, well below the 20% generally regarded as being necessary to ensure secure power supplies.19

PROJECTED UK CAPACITY MARGINS 2012–17

The inescapable conclusion would seem to be that “energy-only” electricity markets will need to be supplemented with “capacity markets” (or markets with capacity obligations) if dispatchable technologies are to remain in the market to provide back-up for when the wind is not blowing at the right speeds.

In the long term, however, nuclear power is likely to be more seriously affected than gas or coal, owing to the higher capital investment costs and hence higher risks in volatile low-price environments.

Some of these effects are already being noticed. Nuclear power in Germany has in recent years moved away from the traditional and efficient role of operating at stable levels close to full capacity, as the introduction of large amounts of variable renewables has repeatedly led to prices below the marginal costs of nuclear, including several instances of negative prices. (Electricity becomes a good of negative value if so much is being generated that costly measures must be taken to prevent overloading the wire system—good news for customers in the short term but disastrous if it results in dispatchable plant being mothballed or not replaced.)

For different reasons (ie that there are times when the available nuclear output is greater than total demand) nuclear power also follows load in France to a considerable extent. Both the French and German experience shows that this can be done technically. Studies suggest that the short-term technological load-following capabilities of nuclear are comparable to coal-fired generation but not as good as CCGT and well behind open cycle gas turbines (OCGT).20 (The overall economics of the last mean that it tends only to be used at times of very high demand.)

The study attempts to quantify the grid-level system costs of various electricity sources in the six countries named earlier. There is quite a degree of variation among these countries, reflecting such factors as the siting of plants with respect to demand, the overall mix, the quality eg of wind cover and the levels of security of supply demanded. Taking the UK as an example, the calculated grid-level system costs of various sources of electricity are as follows. (The figures are cited for a case where the technology in question provides 10% of total electricity and 30% of total electricity and are in US$ per MWh.)

Technology	Nuclear		Coal		Gas		Offshore wind		Onshore wind		Solar
Penetration level	10%	30%	10%	30%	10%	30%	10%	30%	10%	30%	10%	30%
Total grid-level system costs	3.10	2.76	1.34	1.34	0.56	0.56	34.0	45.4	18.6	30.2	57.9	71.7

This implies that introducing variable renewables up to 10% of the total electricity supply will increase per MWh costs, depending on the country, by between 5% and 50%, whereas if the penetration level is 30% this may increase per MWh costs by between 16% and 180% (the last figure referring to solar PV in Finland).

The study also looks at the effect on the profitability of dispatchable technology (and therefore the effect on the incentives for companies to invest in new plant) of having wind and solar at penetration levels of 10% or 30%. (Profitability of these plants is affected both by being taken off line, so losing direct income, and by the very low market price of electricity at times when significant amounts of renewables are available.) The results are striking.

Penetration level		10%		30%
Technology		Wind	Solar	Wind	Solar
Load losses	CCGT	-34%	-26%	-71%	-43%
	Coal	-27%	-28%	-62%	-44%
	Nuclear	-4%	-5%	-20%	-23%
Profitability losses	CCGT	-42%	-31%	-79%	-46%
	Coal	-35%	-30%	-69%	-46%
	Nuclear	-24%	-23%	-55%	-39%
Electricity price variation		-14%	-13%	-33%	-23%

However the system is organised, these costs will need to be met. An economically rational system would place these costs on the renewable technologies themselves, by mandating variable renewables to compete in the market on equal grounds to dispatchable technologies. As noted earlier, in reality renewables are generally shielded by transferring the costs onto the system as a whole, eg through “must-take” contracts that mandate distributors to buy renewable electricity whenever it is available, whatever costs might be incurred on other generators as a result. These market arrangements are just as important, if not more so, as the direct subsidies which have been offered to renewable investors through Renewable Obligation Certificates.

“Smart grids”, which can for example switch fridges or heating systems off for one or two hours when supply drops, may help to mitigate some of these issues to some extent. However, it is difficult to see how a similar approach could revolutionise practice in industry, where cutting power supply would potentially leave a workforce idle. There seem then to be three broad solutions to this challenge.

Capacity payments or markets with capacity obligations, in which variable producers need to buy “adequacy services” from dispatchable providers, which would thus earn additional revenues.

Long-term, fixed-price contract subscribed by governments for guaranteed portions of the output of dispatchable plants whether in the form of contracts for differences or feed-in tariffs.

The gradual phase-out of subsidies to variable renewables, the discontinuation of grid priority and a more direct allocation of additional grid costs to the sources which cause them—this would slow down deployment of renewables (hence reduce costs very considerably) but would also force the internalisation of grid and balancing costs.

The Hidden Greenhouse Gas Emissions of Renewables

Clearly, the need in effect to double the amount of capacity—generating and transmission—to ensure secure supplies in a system with heavy use of variable renewables will result in much higher greenhouse gas emissions (and other resource and environmental impacts) simply from constructing the infrastructure itself. However, the greenhouse gas implications of variable renewables go further. The back-up spinning capacity required to cope with the variability of renewables emits significant amounts of carbon. The inefficiency introduced in fossil generators by having to vary their output to compensate for changes in the output of variable renewables means more emissions of carbon dioxide per unit of electricity generated in these plants. System emissions will therefore be higher than one would believe by simply assuming that when renewables are generating they replace an equivalent amount of carbon-producing capacity. In some cases, at least where coal is the main source of generation, the thermal inefficiency losses outweigh the displacement advantages, resulting in higher emissions than if the wind farms were simply shut down.21 At present these carbon emissions are allocated not to the renewables whose presence causes them but to the back-up capacity itself. This again presents a skewed picture of the true implications of the various sources of electricity.

Conclusions—A More Rational Approach

The NEA report makes four recommendations.

1.Increasing the transparency of power generation market costs at the system level, so it is clear to decision-makers the full economic costs of variable renewables.

2.Preparation of regulatory frameworks that minimise system costs and favour their internalisation—ie the costs fall where they are incurred. Four points need to be addressed here:

the decrease in revenues for operators of dispatchable capacity owing to the compression effect;

the need to internalise the system costs for balancing and maintain supply adequacy effectively;

allocation of costs to the appropriate technology, insofar as it is possible; and

the need for careful monitoring and internalisation of the carbon implications of the requirement for back-up, through a carbon tax again imposed on the causes of the emissions, not necessarily simply the plants that are producing them.

3.Better recognition of the value of dispatchable low-carbon technologies in complementing the introduction of variable renewables.

4.Development of flexibility resources for future low-carbon systems, eg working on increasing the extent to which nuclear plants can follow load, greater storage capacity and increasing international interconnection (a greater challenge for islands like the UK or Japan than it is for a continental nation like Germany or Switzerland).

Whether or not these recommendations are followed, it is vital that the full costs of use of various sources of electricity are taken into account when planning a system that needs to balance costs alongside system security and environmental implications if the best policies are to be followed. The claim that some variable renewable sources are close to “grid parity”—ie are becoming economically competitive with other sources of electricity—tend to be made on the basis that renewables are shielded from their economic and environmental implications. It is of course still a perfectly defensible stance for government to take to argue that certain non-financial advantages of renewables (whatever such advantages might be) merit a very steep increase in power bills or a doubling of the national debt (alongside the much greater land or water area required). But such statements must be made against a realistic assessment of what those costs are and what are the associated reductions, if any, in greenhouse gas emissions from the system as a whole rather than the wind turbine or solar panel taken in isolation. Otherwise we may find consumers paying vastly inflated bills on the basis of promises which cannot be delivered on technically.

30 May 2013