 | |
|
Have the anti-wind power lobby struck again ? A seemingly turbulent researcher from Private Eye magazine rang me on Thursday evening to ask me to revise my interpretation of his “Keeping The Lights On” piece of a few weeks previously. His article seemed at first glance to be quite derogatory regarding the contribution of wind power to the UK’s electricity supply. If I were to look again, I would find out, he was sure, that I was wrong, and he was right. So I have been re-reviewing the annual 2013 “Electricity Capacity Assessment Report” prepared by Ofgem, the UK Government’s Office of Gas and Electricity Markets, an independent National Regulatory Authority. I have tried to be as fair-minded and generous as possible to “Old Sparky” at Private Eye magazine, but a close re-reading of the Ofgem report suggests he is apparently mistaken – wind power is a boon, not a burden (as he seems to claim). In the overview to the Ofgem report, they state, “our assessment suggests that the risks to electricity security of supply over the next six winters have increased since our last report in October 2012. This is due in particular to deterioration in the supply-side outlook. There is also uncertainty over projected reductions in demand.” Neither of these issues can be associated with wind power, which is being deployed at an accelerating rate and so is providing increasing amounts of electricity. The report considers risks to security of the electricity supply, not an evaluation of the actual amounts of power that will be supplied. How are these risks to the security of supply quantified ? There are several metrics provided from Ofgem’s modelling, including :- a. LOLE – Loss of Load Expectation – the average number of hours per year in which electricity supply does not meet electricity demand (if the grid System Operator does not take steps to balance it out). (Note that Ofgem’s definition of LOLE is difference from other people’s “LOLE is often interpreted in the academic literature as representing the probability of disconnections after all mitigation actions available to the System Operator have been exhausted. We consider that a well functioning market should avoid using mitigation actions in [sic] regular basis and as such we interpret LOLE as the probability of having to implement mitigation actions.”) b. EEU – Expected Energy Unserved (or “Un-served”) – the average amount of electricity demand that is not met in a year – a metric that combines both the likelihood and the size of any shortfall. c. Frequency and Duration of Expected Outages – a measure of the risk that an electricity consumer faces of controlled disconnection because supply does not meet demand. The first important thing to note is that the lights are very unlikely to go out. The highest value of LOLE, measured in hours per year is under 20. That’s 20 hours each year. Not 20 days. And this is not anticipated to be 20 days in a row, either. Section 1.11 says “LOLE, as interpreted in this report, is not a measure of the expected number of hours per year in which customers may be disconnected. For a given level of LOLE and EEU, results may come from a large number of small events where demand exceeds supply in principle but that can be managed by National Grid through a set of mitigation actions available to them as System Operator. […] Given the characteristics of the GB system, any shortfall is more likely to take the form of a large number of small events that would not have a direct impact on customers.” Section 2.19 states, “The probabilistic measures of security of supply presented in this report are often misinterpreted. LOLE is the expected number of hours per year in which supply does not meet demand. This does not however mean that customers will be disconnected or that there will be blackouts for that number of hours a year. Most of the time, when available supply is not high enough to meet demand, National Grid may implement mitigation actions to solve the problem without disconnecting any customers. However, the system should be planned to avoid the use of mitigation actions and that is why we measure LOLE ahead of any mitigation actions being used”. And Section 2.20, “LOLE does not necessarily mean disconnections but they do remain a possibility. If the difference between available supply and demand is so large that the mitigation actions are not enough to meet demand then some customers have to be disconnected – this is the controlled disconnections step in Figure 14 above. In this case the [System Operator] SO will disconnect industrial demand before household demand.” And in Section 2.21. “The model output numbers presented here refer to a loss of load of any kind. This could be the sum of several small events (controlled through mitigation actions) or a single large event. As a consequence of the mitigation actions available, the total period of disconnections for a customer will be lower than the value of LOLE.” The report does anticipate that there are risks of large events where the lights could go out, even if only very briefly, for non-emergency customers : “The results may also come from a small number of large events (eg the supply deficit is more than 2 – 3 gigawatts (GW)) where controlled disconnections cannot be avoided.” But in this kind of scenario two very important things would happen. Those with electricity contracts with a clause permitting forced disconnection would lose power. And immediate backup power generation would be called upon to bridge the gap. There are many kinds of electricity generation that can be called on to start up in a supply crisis – some of them becoming operational in minutes, and others in hours. As the report says in Section 2.24 “Each [Distribution Network Operator] DNO ensures it can provide a 20% reduction of its total system demand in four incremental stages (between 4% and 6%), which can be achieved at all times, with or without prior warning, and within 5 minutes of receipt of an instruction from the System Operator. The reduction of a further 20% (40% in total) can be achieved following issue of the appropriate GB System Warning by National Grid within agreed timescales”. It’s all about the need for National Grid to balance the system. Section 2.9 says, “LOLE is not a measure of the expected number of hours per year in which customers may be disconnected. We define LOLE to indicate the number of hours in which the system may need to respond to tight conditions.” The report also rules some potential sources of disruption of supply outside the remit of this particular analysis – see Section 3.17 “There are other reasons why electricity consumers might experience disruptions to supply, which are out of the scope of this assessment and thus not captured by this model, such as: Flexibility : The ability of generators to ramp up in response to rapid increases in demand or decreases in the output of other generators; Insufficient reserve : Unexpected increases in demand or decreases in available capacity in real time which must be managed by the System Operator through procurement and use of reserve capacity; Network outages : Failures on the electricity transmission or distribution networks; Fuel availability : The availability of the fuel used by generators. In particular the security of supplies of natural gas at times of peak electricity demand.” Crucially, the report says there is much uncertainty in their modelling of LOLE and EEU. In Section 2.26, “The LOLE and EEU estimates are just an indication of risk. There is considerable uncertainty around the main variables in the calculation (eg demand, the behaviour of interconnectors etc.)” (Note : interconnectors are electricity supply cables that join the UK to other countries such as Ireland and Holland). Part of the reason for Ofgem’s caveat of uncertainty is the lack of appropriate data. Although they believe they have better modelling of wind power since their 2012 report (see Sections 3.39 to 3.50), there are data sets they believe should be improved. For example, data on Demand Side Response (DSR) – the ability of the National Grid and its larger or aggregated consumers to alter levels of demand on cue (see Sections 4.7 to 4.10 of the document detailing decisions about the methodology). A lack of data has led to certain assumptions being retained, for example, the assumption that there is no relationship between available wind power and periods of high demand – in the winter season (see Section 2.5 and Sections 4.11 to 4.17 of the methodology decisions document). In addition to these uncertainties, the sensitivity cases used in the modelling are known to not accurately reflect the capability of management of the power grid. In the Executive Summary on page 4, the report says, “These sensitivities only illustrate changes in one variable at a time and so do not capture potential mitigating effects, for example of the supply side reacting to higher demand projections.” And in Section 2.16 it says, “Each sensitivity assumes a change in one variable from the Reference Scenario, with all other assumptions being held constant. The purpose of this is to assess the impact of the uncertainty related to each variable in isolation, on the risk measures. Our report is not using scenarios (ie a combination of changes in several variables to reflect alternative worlds or different futures), as this would not allow us to isolate the impact of each variable on the risk measures.” Thus, the numbers that are output by the modelling are perforce illustrative, not definitive. What “Old Sparky” at Private Eye was rattled by in his recent piece was the calculation of Equivalent Firm Capacity (EFC) in the Ofgem report. On page 87, Section 3.55, the Ofgem report defines the “standard measure” EFC as “the amount of capacity that is required to replace the wind capacity to achieve the same level of LOLE”, meaning the amount of always-on generation capacity required to replace the wind capacity to achieve the same level of LOLE. Putting it another way on page 33, in the footnotes for Section 3.29, the report states, “The EFC is the quantity of firm capacity (ie always available) that can be replaced by a certain volume of wind generation to give the same level of security of supply, as measured by LOLE.” Wind power is different from fossil fuel-powered generation as there is a lot of variability in output. Section 1.48 of the report says, “Wind generation capacity is analysed separately given that its outcome in terms of generation availability is much more variable and difficult to predict.” Several of the indicators calculated for the report are connected with the impact of wind on security of the power supply. However, variation in wind power is not the underlying reason for the necessity of this report. Other electricity generation plant has variation in output leading to questions of security of supply. In addition, besides planned plant closures and openings, there are as-yet-unknown factors that could impact overall generation capacity. Section 2.2 reads, “We use a probabilistic approach to assess the uncertainty related to short-term variations in demand and available conventional generation due to outages and wind generation. This is combined with sensitivity analysis to assess the uncertainty related to the evolution of electricity demand and supply due to investment and retirement decisions (ie mothballing, closures) and interconnector flows, among others.” The report examines the possibility that wind power availability could be correlated to winter season peak demand, based on limited available data, and models a “Wind Generation Availability” sensitivity (see Section 3.94 to Section 3.98, especially Figure 64). In Section 3.42 the report says, “For the wind generation availability sensitivity we assume that wind availability decreases at time of high demand. In particular this sensitivity assumes a reduction in the available wind resource for demand levels higher than 92% of the ACS peak demand. The maximum reduction is assumed to be 50% for demand levels higher than 102% of ACS peak demand.” Bear in mind that this is only an assumption. In Appendix 5 “Detailed results tables”, Table 34, Table 35 and Table 37 show how this modelling impacts the calculation of the indicative Equivalent Firm Capacity (EFC) of wind power. In the 2018/2019 timeframe, when there is expected to be a combined wind power capacity of 8405 megawatts (MW) onshore plus 11705 MW offshore = 20110 MW, the EFC for wind power is calculated to be 2546 MW in the “Wind Generation Availability” sensitivity line, which works out at 12.66% of the nameplate capacity of the wind power. Note : 100 divided by 12.66 is 7.88, or a factor of roughly 8. At the earlier 2013/2014 timeframe, when combined wind power capacity is expected to be 3970 + 6235 MW = 10205 MW, and the EFC is at 1624 MW or 15.91% for the “Wind Generation Sensitivity” line. Note : 100 divided by 15.91 = 6.285, or a factor of roughly 6. “Old Sparky” is referring to these factor figures when he says in his piece (see below) :- “[…] For every one megawatt of reliable capacity (eg a coal-fired power Yet he has ignored several caveats given in the report that place these factors in doubt. For example, the sensitivity analysis only varies one factor at a time and does not attempt to model correlated changes in other variables. He has also omitted to consider the relative impacts of change. If he were to contrast his statement with the “Conventional Low Generation Availability” sensitivity line, where wind power EFC in the 2013/2014 timeframe is calculated as a healthy 26.59% or a factor of roughly 4; or 2018/2019 when wind EFC is 19.80% or a factor of roughly 5. Note : The “Conventional Low Generation Availability” sensitivity is drawn from historical conventional generation operating data, as outlined in Sections 3.31 to 3.38. Section 3.36 states, “The Reference Scenario availability is defined as the mean availability of the seven winter estimates. The availability values used for the low (high) availability sensitivities are defined as the mean minus (plus) one standard deviation of the seven winter estimates.” Table 30 and Table 31 show that low conventional generation availability will probably be the largest contribution to energy security uncertainty in the critical 2015/2016 timeframe. The upshot of all of this modelling is that wind power is actually off the hook. Unforeseen alterations in conventional generation capacity are likely to have the largest impact. As the report says in Section 4.21 “The figures indicate that reasonably small changes in conventional generation availability have a material impact on the risk of supply shortfalls. This is most notable in 2015/16, where the estimated LOLE ranges from 0.2 hours per year in the high availability sensitivity to 16 hours per year in the low availability sensitivity, for the Reference Scenario is 2.9 hours per year.” However, Section 1.19 is careful to remind us, “Wind generation, onshore and offshore, is expected to grow rapidly in the period of analysis and especially after 2015/16, rising from around 9GW of installed capacity now to more than 20GW by 2018/19. Given the variability of wind speeds, we estimate that only 17% of this capacity can be counted as firm (ie always available) for security of supply purposes by 2018/19.” This is in the Reference Scenario. The sensitivities modelled in the report are a measure of risk, and do not provide absolute values for any of the output metrics, especially since the calculations are dependent on so many factors, including economic stimulus for the building of new generation plant. Importantly, recent decisions by gas-fired power plant operators to “mothball”, or close down their generation capacity, are inevitably going to matter more than how much exactly we can rely on wind power. Many commentators neglect to make the obvious point that wind power is not being used to replace conventional generation entirely, but to save fossil fuel by reducing the number of hours conventional generators have to run. This is contributing to energy security, by reducing the cost of fossil fuel that needs to be imported. However, the knock-on effect is this is having an impact on the economic viability of these plant because they are not always in use, and so the UK Government is putting in place the “Capacity Mechanism” to make sure that mothballed plant can be put back into use when required, during those becalmed, winter afternoons when power demand is at its peak. Private Eye “Keeping the Lights On” The report from energy regulator Ofgem that sparked headlines on As coal power stations shut down, windfarms are notionally replacing For every one megawatt of reliable capacity (eg a coal-fired power […] In consequence windfarms are being featherbedded – not only with Private Eye page 15 As pandemonium breaks out in newspapers at the prospect of electricity Three years ago incoming coalition ministers were briefed that when Civil servants then devised complex “energy market reforms” (EMR) to make Practical people in the National Grid are now hatching emergency Plan C. Some of these steps are based on techniques previously used in extreme […] ‘Old Sparky’ | |
Categories
Assets not Liabilities Bait & Switch Be Prepared Big Number Big Picture British Sea Power Change Management Delay and Deny Direction of Travel Disturbing Trends Divide & Rule Energy Autonomy Energy Change Energy Denial Energy Insecurity Energy Revival Extreme Weather Fossilised Fuels Green Power Growth Paradigm Hide the Incline Insulation National Energy National Power Optimistic Generation Policy Warfare Political Nightmare Realistic Models Renewable Resource The Data Unqualified Opinion Wind of Fortune