Renewable Energy Foundation

Introduction: Inertia and Frequency
There is much talk about the importance of "energy storage" to enable the adoption of renewables. It is often forgotten in such discussions that the conventional electricity system, of fossil fuelled and nuclear power stations, already has a large storage component built into it. This energy store is found in the rotating mass of the turbine shafts in the generators, and also, to a lesser extent, in the rotating mass of the large electric motors used by some electricity consumers. The rate at which the shafts of those generators, and synchronised motors, are turning is determined by the chosen electricity System Frequency, which in the UK is 50 Hz, or 50 revolutions a second, 3,000 rpm. In almost exactly the same way that a gyroscope has stability and resists attempts to move it due to the energy stored as kinetic energy in its rapidly turning wheel, the synchronised rotations of the electricity generators deliver system "inertia" making it robust against accidents and other surprises, for example an unforecast increase in electricity demand, a grid line failure or the loss of one or more power stations. The energy stored in the spinning mass of the turbines can be drawn down very briefly to buffer the shock and allow time for other generators to increase their output to address the shortfall. In that event, the frequency of the system falls as all the generators slow down due to loss of energy.

Unfortunately, not all generators are capable of operating in this synchronised fashion, and these generators do not contribute to inertia. Solar photovoltaics, for example, have no rotating parts, and wind turbines do not have sufficient mass in their generator shafts to contribute significantly to inertia. Consequently, these generators operate asynchronously, as do the electricity interconnectors with the networks of other countries.

As the proportion of renewable generation and the increased reliance on interconnectors has grown in the UK, the average inertia of the system at any moment has declined, meaning that the system would be less resilient in the face of an accident unless compensating measures were taken, for example the addition of asynchronous compensators (effectively flywheels), generation capable of a very rapid response, such as pumped storage hydropower, or other energy storage devices such as batteries.

It has been assumed hitherto that the UK System Operator, National Grid ESO, was taking adequate steps to ensure that declining inertia was not a threat. However, the load shedding causing local blackouts over the United Kingdom on the afternoon of the 9th of August this year, has put National Grid's management of the system under the spotlight, raising many questions.

The Blackouts on the 9th of August

National Grid has a target frequency of 50 Hz at all times, and is legally required under the terms of its operating licence to maintain frequency between the narrow limits 49.5 Hz and 50.5 Hz. In fact, National Grid's "normal operating limits" are even more stringent at 49.8 Hz to 50.2 Hz. If frequency falls below 48.8 Hz, National Grid will automatically start to disconnect consumers to reduce the demand for energy and try to bring the energy input back into line with energy demand so that frequency can rise to normal levels.

Stable frequency is important for consumers, but it is also critical for synchronised generators. If system frequency falls, this implies that more energy is being taken out than is being generated by power stations. When this happens, those stations that are still connected come under intense mechanical strain, which they cannot tolerate for long. Demand and supply could be thought of as two teams engaged in a tug-of-war. If one of the team members on the supply side suddenly lets go, the strain on the other supply team members increases, perhaps causing injury. The situation facing generators is not dissimilar. In order to protect themselves from major mechanical damage, stations have to disconnect, even though this will almost certainly further reduce the frequency, thus making the system's overall problem even worse. (Generators also have to disconnect if system frequency rises, in other words the turbine shafts start spinning more quickly, because more energy is being put into the grid than is being taken out, though upward frequency excursions, as they care called, are considerably less common.)

The grid event on the 9th of August was a case of an accident in a fragile system leading to power station disconnections ("trips" as they called), and a large fall in frequency placing other power stations under strain, leading to further trips, and then to consumer blackouts to reduce electricity consumption.
National Grid's own interim report shows that the proximal cause of the event was a lightning strike somewhere in Cambridgeshire. This caused some 500 MW of embedded generation to trip. This embedded generation probably included a mixture of wind, solar, and smaller scale fossil-fuelled generators. Just afterwards, or perhaps simultaneously (the interim report is not clear on this point), the Hornsea offshore wind farm reduced its output from 800 MW to 62 MW in a fraction of a second (165 milliseconds to be precise). Shortly after that, the steam turbine at the Little Barford Combined Cycle Gas Turbine power station also tripped. Without the steam turbine it was inevitable that the gas turbines would also have to trip, and this did occur about a minute later.

It is not known why the embedded generators or Hornsea offshore wind were sensitive to the lightning strike. Power systems engineers, have suggested to REF that there may be a feature of the transmission grid in the affected geographical area that makes the frequency protection systems in generators more likely to return a false positive, in other words to trip when they need not have done so. This is as yet uncertain.

Constraint Payments to Hornsea after the Blackout: Saturday–Sunday 10th–11th

Late on Saturday night, the day after the blackout, and for the following 11 hours, National Grid paid Hornsea to reduce output for the first time. This payment was explicitly designated in the market data as resulting from a Constraint, and it appears in our Constraints Payments database.

The volume of energy constrained off was just over 500 MWh for which Hornsea was paid over £98,000. The price charged by Hornsea varied between £164/MWh, and £201/MWh over the period. This variation in price over a short time period is in itself unusual, and deserves investigation by Ofgem, since it seems to suggest that Hornsea was exploiting its position behind a constraint for profit, contrary to the Transmission Constraint Licence Condition. In any case, the price is well in excess of the current market value of Hornsea's Contract for Difference (£158.75). As one of the newer offshore wind farms, its CfD strike price is one of the lowest for offshore wind farms. Consequently, it is surprising that the prices charged in the Balancing Mechanism for reducing output is the highest of all offshore wind farms constrained off this year.

It is not clear why Hornsea was constrained off over this period, but we note that the frequency drifted both above and below National Grid's normal operating limits with a peak at 50.252 Hz and a trough at 49.774 Hz 8 mins 15 secs later, shortly before the constraints to Hornsea began (see Fig 1 below). It is not clear whether the frequency variations, and particularly the excursion below the normal operating limit of 49.8 Hz, is related to the constraint payments made to Hornsea, but so soon after the involvement of this wind farm in a major system failure there is clearly some ground for suspecting that it might be so. The Department of Business, Energy and Industrial Strategy (BEIS) has initiated its own inquiry, alongside those of the regulator, Ofgem, and National Grid. It is to be hoped that these inquiries may shed further light on this question.

Figure 1: The electricity grid frequency at 15 second intervals on the evening of 10 August 2019

Table 1: Constraint payments for Hornsea offshore wind farm on the 10-11 August 2019 for each half hour settlement period where settlement period 48 is for 23:30-00:00 BST or 22:30-23:00 GMT.

At the same time that it was paying Hornsea to reduce output, National Grid was instructing conventional CCGT plant in the region both to increase output and to start generating. These generators included those at Staythorpe, Keadby, South Humber Bank, which all started generating, and Salt End, and Cottam which were instructed to increase output. Thus, instructing Hornsea to reduce output does not appear to be an action intended to address a simple excess of generation in the area but suggests that National Grid may have been concerned at low inertia, and was seeking to increase inertia relative to load by reducing asynchronous generation (Hornsea offshore wind) and simultaneously increasing synchronous generation (CCGT).

We note that National Grid continued to constrain Hornsea Offshore wind on the 15th, 17th, 18th, and 19th of August, curtailing output of 522 MWh, and paying Hornsea just over £100,000 to do so. Prices per MWh remain high, and continue to vary widely, from a minimum of £179/MWh to a maximum of £202/MWh. At the time of writing, 27th August, there have been no further constraints since the 19th of August.

Conclusions

1. Low system inertia is a growing concern in the UK, and there is some ground for thinking that it was at least a contributory factor to the severity of the blackout on the 9th of August. The constraint payments to Hornsea Offshore wind, along with those to CCGT to increase output, are consistent with this hypothesis, and deserve investigation by government and the regulator.
2. We note, that there may also be location specific problems with the network in this area, making the frequency protection systems liable to false positives in the event of a fault such as a lightning strike.
3. In any case, the high and varying prices charged by Hornsea Offshore Wind to reduce output are suggestive of an abuse of market power, and appears to be a contravention of the Transmission Constraint Licence Condition. Ofgem should investigate to determine whether this interpretation is correct, and if it is should intervene to prevent this behaviour and ensure it does not spread into the wider market.

REF wind farm constraint payment data was used on 12th of April 2019, in a Financial Times report on the most recent fault and outage at the new Western Link interconnector. In this post we explore some of the implications of the outage and offer some comments on its probable causes.

A renewable electricity policy based on wind and solar is a grid expansion policy; you can’t have a large fleet of uncontrollably fluctuating renewables in distributed and often remote locations without a cat's cradle of network to connect them and facilitate inter-regional and international exports and imports necessary as part of the demanding system balancing measures required of the System Operator. This fundamental association between wind turbines and wires has been clear for well over a decade, certainly since the now classic warnings of the Eon-Netz wind reports of 2004 and 2005.

But cables, whether subsea, underground, or in the air, are expensive, and add significantly to consumer costs. The Western Link interconnector, 2.25 GW High Voltage Direct Current cable consisting of 385km of subsea cable and 33km of onshore cabling, running from Hunterston in Scotland to Deeside in Wales, required a capital expenditure of somewhat over £1 billion.

Figure 1: The route of the Western Link HVDC interconnector, Hunterson to Deeside. Source: Western Link

 The annual charges added to consumers’ electricity bills permitted by the regulator, Ofgem, to enable the interconnector’s owners to recover this capital and secure a return, are not public knowledge, but are likely to be about 5% of capex per annum, approximately £50 million, for 35 to 45 years. For the owners, National Grid and Scottish Power Transmission, this looks likely to be a gilt-edged investment.

Whether that regulated charge is also good value for the consumer remains to be seen. The Western Link was designed exclusively to provide a channel through which Scottish wind farms could export to English and Welsh consumers. Without that channel, consumers have to pay wind farms to stop generating when local demand within Scotland is insufficient to absorb the wind farm output. Since 2010 when these payments started, up to the present day (12 April 2019), these payments have totalled £562.5 million, with £124.6 million of that being paid in 2018 alone. Payments so far in 2019 amount to £52 million. For the short periods when the Western Link has been running since coming online at the end of 2017, two years late, the interconnector does seem to reduce constraint payments (see Figure 2 below), but no consistent picture is emerging because the Western Link itself has proved extremely unreliable.

The history can be summarised thus:

[1]  The interconnector was expected to be ready in late 2015, but did not go live until December 2017, two years late, and only then at half capacity.
[2]  It was expected to become fully operational in June 2018. However, the link experienced problems, and tripped soon after it was turned on after the repairs in May. It underwent further repairs, still with a target of commissioning in June.
[3]  During testing in June the Link tripped again, and it became clear that this was no minor teething problem. The company offered no predicted timetable for a return to service.
[4]  At the end of July 2018 Western Link announced that it would be operational in September.
[5]  On the 12th of September Western Link revealed that they had experienced another fault resulting in a trip and that consequently the September target would not be met.
[6]  On the 16th of October the Western Link returned to service, and worked apparently without incident until the 19th of February 2019 when it tripped again.
[7]  This fault was repaired in a month and the link re-opened on the 22nd of March. However, on the 6th of April the Western Link failed once more.
[8]  The company has revealed that the latest problem appears to affect a subsea section of the cable, some 150km from Hunterston.The company is predicting, perhaps optimistically, completion of repairs by the end of May.

All projects have teething problems, but two years late and five service failures in the initial phase of operation is beginning to suggest a systemic problem with the project. Unfortunately, due to a dearth of information in the public domain, analysts can do no more than speculate about the underlying causes. At present much of the attention is focused on the manufacturer of the cable, Prysmian, which has, as the FT story reports, said that its earnings may be reduced by €80 million because of the problems with the Western Link. Whether that means Prysmian is expecting to take sole responsibility is unclear, since its partner in the construction, Siemens has yet to issue a statement, and neither of the owners National Grid or Scottish Power Transmission is saying anything beyond the bare-bones statements on the Western Link website.

Clarity on this point is a matter of public interest, since the late delivery and the failures of the Link will have resulted in significant extra consumer costs, comprising a) Additional constraint payments to wind power to stop generating, as well as b) Payments to conventional generation south of the constraint to bring the system back into balance.

These latter payments can only be confidently identified by the System Operator, National Grid, but we can calculate the constraint payments paid to Scottish wind farms via the Balancing Mechanism during the periods when the interconnector should have been operational. The following table, uses outage information from the Western Link site and provides wind constraint payment data and constrained electrical energy volume data from the REF wind constraints archive.

Table 1: Western Link Service History, 2016–2019, with associated wind power constraint payments and volumes of electrical energy. Source: Western Link: REF constraint payment database, calculations from Balancing Mechanism Reporting Service (BMRS) data.

Figure 2: (a) The available capacity of the Western Link interconnector from 2016 to 13 April 2019 showing outages and reduced capacity; and (b) the constraints costs in £million per number of days in the relevant period paid to Scottish wind farms. The numbers on the plots refer to the key in Table 1.

Over the period 2016–7th December 2017 the interconnector was unavailable due to late delivery, and constraint payments to wind in Scotland totalled £168m. During the period December 2017 to 5th May 2018 the interconnector was running at reduced capacity and constraints totalled £25m. Payments during the outage from May to October 2018 amounted to £74m, and during the outage 19 February to 22 March 2019 a further £30m. Adding this all together, we see that £168m was paid over the late delivery period, and during the outages and reduced capacity periods, a further £129m. Furthermore, as indicated above, it should not be forgotten that there are also payments to conventional generation south of the grid constraint to re-balance the system.

Not all of these costs would have been avoided if the interconnector had been working as promised, but some considerable part would have been prevented. It has been reported that Ofgem is now investigating the Western Link failures to determine if the late delivery and outages are to be blamed on gremlins, or result from negligence. Ofgem will also presumably be asking whether some part of the resulting constraint payments on both sides of the bottleneck, can be attributed to that negligence, if any, and the consumer costs recovered from the responsible party or parties.

The three areas that Ofgem will probably be examining are:

1. Failure of quality control in manufacturing
2. Faulty installation
3. Inappropriate specification of the equipment

Of these, the first two would appear to be the responsibility of Prysmian and perhaps Siemens, as the manufacturer and installers. The question of specification, however, could also involve both National Grid and Scottish Power Transmission as commissioning owners. Such an error of specification, if not negligent, would be understandable; honest mistakes of this kind are not uncommon in large projects. However, there must be a suspicion that the specification of the project was deliberately and over-ambitiously engineered to reduce costs at the hazard of reliability. Such a trade-off would be unsurprising in this case, since transmission assets designed solely to carry wind energy — and the Western Link has no other purpose — are very likely to be under-utilised, due to the low load factor of wind farms, which onshore rarely exceed 30% over a year. This problem is endemic in the renewables industry, and is well known in the offshore wind industry, for example, where the attempt to drive down costs and preserve market position as subsidies are scaled down and cancelled, has resulted in a higher rate of cable failures. Indeed, the FT quotes the consultancy 4C offshore to the effect that there have been 40 cable failures at offshore wind farms since 2014, and other sources suggest that 90% of offshore wind farms have had problems with their cables, with the blame being laid squarely at the pressure from developers for cost reduction, with manufacturers responding to pressure by introducing new and relatively untested products to meet customer specifications (see “High Price of Low Cable Costs” and related stories, ReNews (7 February 2019), 11–13).

This could be true of the Western Link, which is a novel kind of interconnector, being the first such project in the world to transmit DC power at 600kV, rather than the more normal 500kV. The increase in voltage is made possible through the use of a novel form of insulation, Paper Polypropylene Laminate insulation impregnated with fluid (see “Ready, Willing and Cable”, Power Engineering International 5 Jan. 2018. The beneficial economic consequences of the higher voltage made possible by this new insulation system are a parallel 20% increase in the power capacity of the cable and a reduction of transmission losses by a third. For an underutilised cable these are very tempting prizes, and it is quite plausible that National Grid, Scottish Power and their contractors decided to take a risk on a relatively untried technology. If that is the case, the gamble is not turning out well.

We are left, then, with a question: Is the fault-prone state of the Western Link interconnector the inevitable outcome of inorganic, policy-driven, coerced and overheated sectoral growth resulting in advanced technological deployment well ahead of the learning curve and at the expense of the consumer? If the answer to that question is Yes!, government will have to admit that it too has a share in the blame.

Dr Mark Carney, Governor of the Bank of England, has recently called for government to cease using to the widely discredited Retail Prices Index (RPI), for example when index-linking to offset inflation. This follows a recent House of Lords report on the indices used to measure consumer price inflation which drew attention to the fact that due to an error in the methodology used by the UK Statistics Authority (UKSA) the RPI is greater than the Consumer Prices Index (CPI), with the difference tending to increase (see Figure 1).

As the House of Lords report states, this error has created both winners and losers. Examples of losers include students and commuters whose loans and rail-fares are linked to the higher and more rapidly increasing RPI. Amongst the winners are holders of RPI-linked gilts and also, recipients of renewable energy subsidies under the Renewables Obligation (RO), the Feed-in-Tariff (FiT) and early adopters of the Renewable Heat Incentive (RHI). Since the RPI is used to calculate the inflation proofing of these subsidies there is now significant and unjustifiable over-support, with the scale of the problem being particularly significant in the largest of the schemes, the Renewables Obligation.

Figure 1: The Retail Prices Index (RPI) and the Consumer Prices Index (CPI), from 2010 to 2018. Source: ONS.  Ofgem sets the RO buy-out price by taking the buy-out price from the previous obligation period and adjusting it in line with the change in RPI for the previous calendar year. E.g. For the obligation period 2016-17 the price was set at £44.77 per ROC  an increase of 1% (2015 RPI)  from the 2015-16 value of £44.33.

However, the most recent of the renewables subsidies, the Contracts for Difference, is indexed against the Consumer Price Index, as have those registering for the RHI since April 2016, suggesting that government is now aware of the excessive annual uplift provided by the RPI linkage and recognises that it should no longer be used for this purpose. This is to be welcomed, but it is difficult to see why government ever accepted the case for RPI index linking for the Renewables Obligation. A significant part of the difference between RPI and CPI arises from the treatment of housing costs in the RPI, and the cost of housing has no bearing on the cost of generating renewable electricity, making the RPI an obviously unsuitable index for the purpose of inflation proofing subsidies to that sector. A second component of the difference is the way in which price changes are weighted. The RPI uses a method that is quite inappropriate for companies that have incurred most of their costs during the construction phase, as is the case with renewable energy projects. Overall, the use of RPI does not reflect either the costs or the investment incentives appropriate to renewables developments. It was simply the wrong choice, as government implicitly admits by using the CPI for the Contracts for Difference.

However government has not yet acted to correct its error in relation to the Renewables Obligation and other schemes. This is particularly puzzling since in other instances, government acted promptly. For example, in 2011 government changed the index-linking of pensions and other welfare benefits from the RPI to the CPI. Why did it not at the same time apply this change to the index linking of the Renewables Obligation electricity subsidy where consumers foot the subsidy bill? This failure in relation to the RO has, we estimate, already needlessly cost consumers over £1 billion and we estimate will cost a total of at least £9 billion up to the end of the Renewables Obligation in 2037. £6 billion of this excess subsidy would go to wind power generators. It is obviously wrong, indeed iniquitous, that investors in renewable generation should be treated so much more favourably than pensioners and those on benefits.

To put this in technical terms, continuing to use RPI to adjust the RO buyout price means an expected subsidy of over £80/MWh generated in 2036-2037 by comparison with £64/MWh had the Government switched from RPI to CPI in 2011.

It is not too late for government to act, and, at a minimum, if the switch to CPI is implemented now to the Renewables Obligation, we estimate that this measure alone would save consumers at least £3.6 billion in undeserved subsidy payments over the remaining lifetime of the Obligation.

However, the detrimental effect on consumers is so significant that we believe that the government should go further and cut the RO buyout price to the level it would now be at if CPI indexation had been applied from 2011.

Such a move would reduce the buy-out price for the 2019-2020 period from £48.78 to £45.31 per MWh. If the Government predictions for the size of the Renewables Obligation prove accurate, this change would save consumers around £400 million in subsidy for this one year alone, and about £7.7 billion in total up to 2037.

About one third of these savings would affect domestic consumer electricity bills directly, while the other two thirds would benefit households indirectly through the reduced costs of goods and services provided by industrial, commercial and other consumers. It would also benefit households by alleviating the downward pressure exerted by the Renewables Obligation on wages and on rates of employment.

The renewables industry would doubtless protest at such reforms, but the subsidies are already generous, delivering a rapid return on capital, and the industry can have had no reasonable expectation that these subsidies would be indexed at a rate above that of inflation.

In addition to reforms of the Renewables Obligation, government should also investigate the possibility of reforms to the index-linking of the Feed-in Tariff and further retrospective reforms to the Renewable Heat Incentive.

It is a commonplace that the uncontrollable variability of the most prominent modern renewables, solar and wind power, requires electricity storage to make them viable as free-standing and independent technologies. Consequently, batteries and other storage systems are sometimes presented as the key to the problem of intermittency, storing green energy at times of surplus for those times of dearth when the wind isn’t blowing or the sun shining, and so unlocking the wholly renewable future. Batteries are now on the verge of broad deployment in the UK, but they do not appear to be taking up the function expected of them, and instead seem to be gravitating towards close working with conventional generation.

The UK government, amongst others, has taken a close interest in the encouragement of energy storage research and deployment, sponsoring a competition to reduce the cost of energy storage technologies and more recently proposing a revision to the planning regulations to make it easier for developers to obtain consent for smaller proposals where storage is combined with a generation asset. (BEIS, Consultation on Proposals Regarding the Planning System for Electricity Storage (January 2019) 

This reform, which aims to reduce the burden on the Nationally Significant Infrastructure Project (NSIP) system by directing projects towards local authorities, has clearly been prompted by the strength of interest in developing electricity storage projects in general and combined generation+storage projects in particular. The rapid growth in that interest can be gauged by examination of planning system data now being released by the Department of Business, Energy and Industrial Strategy, as part of its Renewable Energy Planning Database, a choice of venue that clearly indicates the sector in which government expects storage to have most relevance. The following charts, generated by REF from this data, provide an overview.

Figure 1: Electricity energy storage sites by technology type and status in the planning system. Chart (a) shows the installed peak discharge capacity in MW in each category and (b) the number of sites.

The operational fleet is dominated by long-established conventional pumped storage (Cruachan, Dinorwig, Ffestioniog and Foyers), with a major contribution from the 800 tonne flywheels used to provide surge power for the Joint European Torus (JET) fusion project in Culham. As yet, there is relatively little operational battery storage, with only twenty-nine projects totalling a total peak output capacity of somewhat over 300 MW.

However, there are already 182 battery proposals either consented and awaiting construction or in the planning system and seeking permission. The majority of those projects are below 50 MW in peak output, but the aggregate total is about 3 GW, slightly exceeding proposals for new pumped storage.

Things are obviously beginning to move fast for batteries, and two operational projects are already visible in the Balancing Mechanism (BM), Arenko’s Bloxwich scheme a battery with a peak output of 41 MW, and Centrica’s Roosecote, a 49 MW project at the former gas-fired power station site in Barrow-in-Furness. (In passing it should be noted that in the charts and commentary above we have corrected the omission of the Bloxwich scheme from the government’s planning data on energy storage, as well as errors in the entry relating to the Coire Glas pumped hydro scheme, but this raises the possibility that there may well be other significant errors and omissions in the government’s data; this is clearly work in progress.)

Many of the battery schemes currently in development are built alongside generation, and some 63 are, to use the jargon, “co-located” in this way. What will, perhaps, be surprising to many is that 20 of these schemes, with a peak capacity of 563 MW, are in fact co-located with fossil fuel generation, not renewables. The following figure charts the relevant data, showing batteries described as standalone, co-located with fossil fuel generation and with renewable energy:

Figure 2. Breakdown of large scale battery storage in the planning system showing difference between stand-alone installations compared with those co-located with conventional fossil fuel plant or renewable energy – typically onshore wind or solar PV.

A prominent example of this tendency can be found at Tilbury Power Station (now rebranded as the Tilbury Energy Centre), where RWE has planning consent to build a battery with a peak output of 100 MW. Other elements in the Tilbury proposal include a 2.5 GW Combined Cycle Gas Turbine, and a 299 MW Open Cycle Gas Turbine (OCGT).  While standalone battery projects dwarf co-located schemes, batteries co-located with fossil fuels currently dwarf co-location with renewables.

There are several reasons making co-location at fossil fuel sites highly attractive. With an existing grid connection, and a skilled workforce, the addition of another revenue earning asset on the site may well be a straightforwardly sensible move. Furthermore, the presence of a large renewable generating fleet has meant that electricity demand on the Transmission System is becoming increasingly stochastic (random), with low inertia making system frequency excursions more likely, and the System Operator is consequently finding it much harder to balance demand and supply and to maintain power quality. As a result, favourable contracts are available to generators able to offer very rapid response to requests for grid balancing services, extra generation at very short notice amongst them. Fossil fuel generators can respond quickly, within an hour, and some much more quickly than that, but even reciprocating diesel engines and OCGTs need minutes or at best several seconds to react, and the System Operator would ideally like a response that is almost instantaneous. Batteries can provide such super-rapid services, in a fraction of a second, but are not ideally suited, largely because of cost, to sustaining the service over a longer period if required. Consequently, it makes sense to combine a battery and a conventional generator, with the battery providing the first phase of that response for seconds and minutes until the co-located fossil generator can ramp up and take over. Since the fossil fuel generator can operate at will, the battery can easily and inexpensively be kept fully charged.

This close combination is most readily achieved by physical co-location, as described in the government’s data, but virtual combination between battery and fossil fuel generation is also entirely feasible, leading us to suspect that some, at least, and perhaps many of the stand-alone battery projects are also operating in tandem with conventional generators.

None of this should provoke either surprise or outrage. Given the need to address the emergent balancing and power quality problems on the UK system, mostly caused by renewables, the real or virtual co-location of electricity storage and conventional generators is perfectly rational engineering, and probably makes the best of a bad job for the consumer. But it is not at all the role for batteries that many, perhaps even government, will have expected. Rather than storing green energy to smooth wind and solar output over longer periods, leading to a future where renewables become independent of conventional generation support, many of the UK’s batteries will actually be storing fossil electricity to enhance the ability of conventional fossil-fuelled generation to react to the difficulties arising from wind and solar. That independent future for renewables may be a little closer, but it is still very distant.

Constraint payments to wind farms, that is payments to stop generation, mostly in Scotland, reached record levels in 2018, with the total reaching £124,649,106, as compared to the total in 2017, of £108,247,860.

Of this, £115,716,335 was paid to Scottish wind farms, and nearly all of that, £115,313,091 went to onshore wind farms. These costs are, of course, passed through to consumers in their bills.

The new record data generated a number of stories in the press including the Sunday Times, the Times and the Scottish Daily Mail.

Of particular interest is that behind this record lies the fact that many wind farms received constraint payments for the first time in 2017 and 2018, as shown in the map below, including some such as Stronelairg that began to be constrained off (28 December 2018) within weeks of being connected to the system.

Figure 1: Newly constrained wind farms in 2017 (Green dots) and 2018 (Red dots)

There is a growing suspicion that the probability of major constraints is a factor in site selection, since it increases the average earnings per MWh generated as a result of the scale of compensatory constraint payments allowed, which are, and we think unjustifiably, well above lost income. Constraints can account for a substantial part of the potential output of a wind farm.

It is possible from market data to produce reasoned estimates of the fraction of output that is discarded. The following table lists the wind farms with the highest proportion of constrained off energy, which ranges from just below 30% to around 15%. These are high fractions, and, given that constraint payment compensation, which averages £70/MWh, is over 50% in excess of the lost income (£45/MWh) make a material difference to the annual income of the site.

These matters are not at present adequately addressed in the planning system, but clearly should be. Decisions makers should be aware that a site may have been chosen precisely because it lies behind a constraint, and in spite of other material considerations, such as local environmental impacts.

Table 1. Estimated proportion of total generation in 2018 that received payments through the Balancing Mechanism not to be generated. For example, Bhlaraidh generated approximately 192 GWh in 2018 and a further 80 GWh was constrained resulting in the 29% figure in Table 1.

The long overdue commissioning of the WesternLink Anglo-Scottish subsea interconnector, completed earlier this month, appears to be mitigating the need to constrain off Scottish wind power, but the delay in the cable’s delivery has meant that it has been overtaken by overall growth in wind on the network, and constraint payments continue, with over £2m being paid out to forty wind farms yesterday, the 23rd of October 2018, with an average price of £69/MWh, well above the lost income of £45/MWh.

Nevertheless, the WesternLink does appear to be facilitating very high, record-breaking levels of wind on GB’s electricity system. For example about 11 GW of wind was carried over a single settlement period (SP 41 which covers the period 8:00- 8:30 PM) on the 23rd of October 2018. This can be seen clearly in the two figures below, generated from REF’s free, online fuel mix database, which is based on Nationall Grid data.  The first panel shows the total fuel mix, and the second panel the renewables and interconnectors alone.

Nevertheless, in spite of the high levels of wind actually facilitated, constraint payments are still very significant, particularly between midnight and 8am, when demand is low. For instance, in Settlement Period 1 on the 23rd, just after midnight, GB transmission system demand stood at 22 GW, while total wind was approximately 10.8 GW, of which 8.4 GW was actually used, with some 2.4 GW constrained off at a cost of £86,000 for that half hour period. Without constraints there would have been about 50% wind on the system.

It is interesting to note that the interconnectors to France and Holland were acting as import channels at this time, with a total of 1,660 MW, a reminder that the system and its markets are not, and probably never can be, made infinitely flexible in the interest of accommodating wind.

National Grid had earlier predicted a Balancing Services Use of System (BSUoS) cost of £2.11/MWh for the 23 October 2018, and it will be interesting to see if these forecasts, which reflect expectations for the WesternLink, are accurate.

Constraint payments to wind power are hitting new records on a regular basis. The highest daily total, £4.77m occurred on the 8th of October 2018, and the highest monthly total of £28.4m in September 2018, a staggering £5m more than the previous record of £23.2m in October 2017. The annual record, set last year, of £108m looks almost certain to be broken this year, where the total is already £101.5m to the 19th of October 2018.

Constraint payments to wind power, mostly but not now entirely in Scotland, comprise a staggering 8% of the cost recovered through the Balancing Services Use of System (BSUoS) charges, with a very substantial proportion of the remainder being caused by wind constraints that require conventional generation to be constrained on to the system south of the constraint to make up for the absence of contracted wind.

Some part of these records are the result of the late delivery of the 2,250 MW WesternLink High Voltage Direct Current (HVDC) link between Hunterston and Deeside, which was intended to enter service at the end of 2015, but has only just been commissioned in September 2018, due to a series of faults that must be both embarrassing for its builders, Siemens and Prysmian, and financially disappointing for its owners, National Grid and ScottishPower Transmission.

But the late arrival of this very expensive, more than £1bn, sticking plaster, probably adding upwards of £50m a year to consumer bills (it is a rule of thumb that grid imposes a standing charge on the consumer of about 5% of the capex for the 30 to 50 life of the asset), cannot explain all the increase observed, and certainly cannot provide a complete solution to what is clearly an acute and growing problem for the system.

What is going on? National Grid is being fairly cagey, and has not yet released comments on the vast constraints paid in September, but there is information coming out about the significant costs during the weekend of the 28th and 29th of July, when over £7m was paid out, with one large offshore wind farm in English and Welsh waters, alone, receiving about £1m over the period, according to the Balancing Mechanism data that REF publishes.

Presentations delivered at National Grid’s “Transmission Operation Forum” show that this was an event novel in character, consisting both of the by now familiar combination of high wind output and low demand (renewables are poorly correlated with demand patterns), and a series of unfortunate coincidences on the network. National Grid describes the event in the company’s obscure, acronym and jargon-laden Powerpointese thus:

There were several limitations on system operation for the ESO [Electricity System Operator], across both System and Energy. The flow of power North to South was restricted by a significant year ahead outage coupled with the HVDC [High Voltage Direct Current interconnector –the Western Link] not yet commissioned. In addition, a lack of conventional generation, displaced by high wind output, resulted in Negative Reserve, Voltage and ROCOF [Rates of Change of Frequency] problems. This resulted in actions on high priced wind units to solve Negative Reserve and Response requirements.

We can try to put this more clearly: the flow of energy from wind power in the North was restricted because some grid lines were out of action due to long-planned maintenance, a situation that was made worse by the fact that the WesternLink, which was expected to be in service by now, had experienced yet another fault. This basic difficulty was compounded by the fact that conventional generation, probably gas fired power stations, had been displaced from the market by wind power, leaving the grid network vulnerable to problems caused by voltage fluctuations, rapid changes in system frequency (which risks tripping embedded generators off the system causing cascading problems), and concerns that they would not be able to call on sufficient generation to reduce output and contain upward excursions in system frequency (negative reserve).

The high cost in dealing with this set of problems resulted from a novel development: the constraint boundary, normally located on the Anglo-Scottish border moved down to a location in the midlands, leaving several large offshore wind farms, including West of Duddon Sands, near Barrow-in-Furness, north of the constraint. Offshore wind farms charge more to reduce output than onshore wind farms ostensibly because they lose more subsidy per MWh lost when constrained off. As a matter of fact, and as shown in the REF blog in July, the charges ranged from £28 to £79 per MWh in excess of the subsidy lost.

National Grid illustrates the 28–29th July 2018 weekend problems with this map:

What do we learn from all this? As has long been predicted by systems analysts and grid engineers with practical experience of systems operation, the presence of large volumes of renewables on a system such as that of the UK will very significantly increase its fragility, making it vulnerable to unfortunate coincidences of adverse circumstances, such as those on the 28th and 29th July 2018. Addressing these problems is not, at least at present, impossible, but it is very expensive, and becoming more so.

We now await with great interest National Grid’s detailed explanations of the problems that required them to spend £28m of consumer funds on wind constraints in the month of September 2018 and the eye-watering £12.5m spent in just three days over 7–9th October 2018. Are these events exceptional, or just the New Normal?

September 2018 has seen the highest monthly payments to wind farms to stop generating since records began in 2010.

Constraint payments to wind farms in the United Kingdom totalled £7.12m over the weekend, 28–29 July 2018, making it the most expensive weekend to date and well above the previous record of £5.87m for 24-25 June 2017. Constraints on Saturday the 28th of July amounted to £4.41m, and on Sunday the 29th to £2.71m.

Over the last few days REF has been delving into the wind constraint payments data to assist an investigation by the Scotsman newspaper into ongoing problems with the Western Link HVDC Interconnector, a 2.2 GW, £1 billion subsea cable from Hunterston to Deeside expressly built to carry Scottish renewable electricity to English and Welsh consumers.