Case Study on Methods of Industrial-scale Wind Power Analysis (Part I)

By Kent Hawkins -- April 6, 2010

Is the introduction of industrial- or utility-scale wind power into our electricity systems good public policy?

This political economy question (wind power is government dependent, or it would only be a market question) hinges to a large degree on operations research, or engineering. And it is here that a hotly contested debate is going on, for it is an open question about how much wind power really displaces fossil fuels–the raison d’etre of wind subsidies in the first place.

This two-part series evaluates some of the latest approaches and considerations in this debate. One important paper published in 2009 by Charles Komanoff sees wind-for-fossil-fuel displacement as robust and is currently being cited by wind proponents in Maine. Another paper in my review is a study by Gross et al, which is relied on by Komanoff.

Part I critically evaluates Komanoff by extending the critique of Milligan et al; Part II focuses on the important consideration of electricity generation capacity value and analyzes some of aspects of the referenced Gross et al paper.

Introduction

The following is a summary of important areas of consideration in assessing documents such as Komanoff’s:

· Treatment of wind volatility

· Use of emotive and pejorative language in referring to those of an opposing opinion. On the other hand, those with the same views are treated with complimentary descriptions. This treatment raises red flags that invite closer analysis.

· Distinction made between very small and larger wind penetrations

· The realities of wind production in Denmark.

· Complete consideration of the impact on overall system capacity requirements, fossil fuel consumption and CO₂ emissions in connection with the displacing of some fossil fuel plant production with highly intermittent renewables.

· Evaluation of normal operating reserves as sufficient to act as wind shadowing/backup .

· Assumptions about the fast ramping capability of nuclear and coal-fired plants

· Assessment of wind’s capacity value

· Dependence upon a report by Gross et al, which itself is not convincing.

· Questionable arithmetic and graphical representation approaches

· Claims about the benefits of geographic diversity and the possibility of improved wind forecasting

Wind Volatility

At the beginning Komanoff quotes what he wrote a few years ago:

“[S]ince wind is variable, individual wind turbines can’t be counted on to produce on demand, so the power grid can’t necessarily retire fossil fuel generators at the same rate as it takes on windmills. The coal- and oil-fired generators will still need to be there, waiting for a windless day. But when the wind blows, those generators can spin down.”

The underlying assumption, which appears to be the basis for most of the following arguments, is that when the wind blows it does so steadily, and when it ceases it remains so for a day (or more). Unfortunately wind does not blow steadily and may not always cease for long periods, and this is at the heart of the issue, which can be summarized as follows:

• On days of little or no wind (electricity is not produced at wind speeds less than about 9 mph), wind plants make their most useful contribution of any period by allowing the other generation plants to be utilized in their normal mode of operation, which is their most efficient in terms of fossil fuel use and CO2 emissions.
• On days when wind is available, even at moderate speeds in excess of 9 mph, significant variations can occur over the period of a fraction of an hour, ranging over the full scale of wind plant capacity. The electricity output variation of wind turbines is related to the cube of the wind speed, that is, any change in wind speed by a factor of two is magnified by eight times (and 3 times the wind speed change produces 27 times the change in electricity output). As electricity users require a steady, reliable supply, wind output must be “mirrored” or “shadowed” by conventional generation means to render it useful.

The point is that wind output volatility is a major factor in any discussion about the integration of wind power in an electricity system. In the absence of sufficient impounded hydro capability, which is typically the case, different plant types are required (e.g. increased use of OCGT plants, which are more responsive but less efficient than CCGT plants) and those involved are forced to operate at a lower level of average production, increasing their costs per megawatt-hour (MWh). Of greater importance, as they are being used in a more inefficient mode than normal operations, they consume more fossil fuel and produce more CO2 emissions per MWh. This is like a car in continuous stop/start speed-up/slow-down city traffic as opposed to steady highway driving. This mode of operation can totally offset any reductions claimed in fossil fuel consumption and CO2 emissions as a result of the presence of wind.

Use of Language

Appropriate treatment of a subject should avoid questionable and unbalanced language in describing either the author’s views or those of opponents. Table 1 supplies examples from Komanoff`s paper.

Table 1 – Examples of Questionable Language Used

Effect of Wind Penetration

One of the main observations on Komanoff’s paper is that the majority of the conclusions are based on electricity systems with very small wind penetration. At this level, the effect of wind’s volatility is generally not significant enough to cause noticeable effects. As wind penetrations are increased above about 1-2 per cent, in energy (MWh) terms, the problems cannot be ignored. This is confirmed by experience in Germany, Denmark, Spain, Texas, U.S. Pacific Northwest, Alberta and Ireland, for example. To say that problems may be eventually solvable ignores economic realities, as essentially everything is solvable if enough money is provided. The real-world costs for solving these issues are typically not completely considered.

Significant emphasis (about one-third of Komanoff’s paper) is placed on the experience of the PJM grid in Pennsylvania, which has a penetration of all renewables (of which wind is a portion and excluding hydro) of only 1.1 per cent in 2007.

Denmark

Early in the paper Komanoff introduces the almost always misunderstood notion that Denmark’s domestic use of wind-generated electricity is 20 per cent. Although Denmark does produce somewhat less than this amount, it exports most of it at very low prices largely to Norway and Sweden. Why does Denmark do this? It is because its electricity system cannot withstand this level of volatile production and must discard wind output above as little as about 5-6 percent of the total electricity system production. A Danish energy consulting firm reports that a country’s fleet of wind plants can operate like a single, virtual out-of-control power plant.[1] Claims that this reduces fossil fuel use and CO2 emissions elsewhere are questionable. In Norway/Sweden it is balanced by hydro plants and displaces some of their production, especially in dry seasons.

The claim that the electrical energy thus “stored” in Norway’s and Sweden’s hydro system and later retrieved (at higher costs) can be countered with: Denmark would likely be further ahead by eliminating its wind plants, (including wind generation’s high costs and two way transmission losses), and buying the hydro-produced electricity when needed.

Fossil Fuel Plant Displacement

Komanoff`s point that wind need not displace fossil fuel plants, but the production from them, is reasonable, although the notion appears to be held to some degree by wind proponents that coal plant elimination (existing or planned) is the expectation. Wind may replace fossil fuel electricity production, but as already indicated, the extension to fossil fuel consumption and emissions savings is not necessarily the case.

Another interesting observation is that, as wind penetration approaches the level in Germany, the introduction of wind increases the total system capacity needed to satisfy a given level of demand by almost the total wind capacity. This has been clearly stated by E.ON Netz[2] and confirmed by Hoppe-Klipper.[3] A representation of the referenced slide in the Hoppe-Klipper presentation and the above-referenced E.ON Netz report is shown in Figure 1 and illustrates this duplication of capacity.

Figure 1 – Redundancy of Wind Power in Germany

The first column in Figure 1 is the demand for 2007. The second, shown in the darker blue, is the conventional generation capacity to meet the 2007 demand including reserves. The third column is the effect of significant wind presence, shown as lighter blue, and illustrates the increased total electricity system capacity as a result. The increase is almost the total wind capacity. Subsequent years show projected growth in demand and wind capacity. The substantial growth in wind has no significant increased impact on reducing other generation means and continues to add to the total capacity requirements by almost the total wind capacity. This effect is due to the diminishing capacity credit for wind as its penetration increases to the 2015 and 2020 levels for Germany as shown in Table 2, which is based on information from the sources given above.

Table 2 – Redundancy of Wind Power in Germany

 

Operating Reserves

Komanoff devotes about 3 (out of 10) pages to a discussion of reserves based on the PJM grid system, which, as already stated, has a very low penetration of wind power. Based on PJM’s actual experience, the impact of this minimal amount of wind on the “supplementary reserve” component, which is called up when the immediate synchronized reserve component is overused (PJM terminology) is itself minimal. This may be consistent with the experience in jurisdictions with very low wind penetrations, where normal reserves may be reasonably sufficient to balance the wind volatility, along with the fluctuations in demand, without jeopardizing system reliability.

Komanoff’s discussion includes the following quote from the PJM chief, Karl Pfirrmann:

“Wind is not as variable as people may think. Our experience shows that, if a wind generator is operating at a certain level at present, there is an 80 percent probability that it will be operating within ±10 percent of that level one hour from now. And, there is a 60 percent probability that it will be operating within ±10 percent of that level five hours from now. We’re also encouraged that better forecasting will enable us to better predict the output from the wind generators on our system.”

This displays misunderstanding of the performance of wind plants, especially during periods of high production. As already indicated, within the course of each hour the wind output can fluctuate between almost 0-100 per cent of full capacity, to say nothing about a five hour period. It could well be the case that the level of wind output at hour intervals will not display this high degree of volatility. The quote from Pfirrmann by no means supports Komanoff`s conclusion that “This suggests that rapid changes in wind output requiring rapid turndown or ramp-up of fossil plants on the grid are relatively infrequent…”.

Even if wind could be forecast with 100 per cent accuracy over 5 minute intervals, this does not change the need for the mirroring/shadowing backup to have to follow the high level of volatility, which is the major source of the problem with respect to the consumption of fossil fuel and production of CO2 emissions, not forecast accuracy.

Pfirrmann goes on to say that in connection with the impact of wind’s volatility on the synchronized reserve, which serves as protection against a sudden loss of the single largest generating unit on the entire system, and the amount they (PJM) maintain is based solely on the size of that [sic] largest generating unit, Komanoff concludes:

“Since the largest generating unit on almost every U.S. grid is on the order of 1,000 megawatts, whereas individual wind turbines are only several megawatts and even entire wind farms are rarely more than several hundred megawatts, it’s clear that wind power imposes no additional synchronized reserve requirements on power grids.”

The 1,000 megawatts is a reasonable size for this consideration, but to compare this to the loss of output from a single wind plant of a few hundred megawatts does not stand close scrutiny. If the wind plant failure was an equipment failure, then the conclusion applies. However on a relatively frequent basis (even intra-day) it is not unusual for a fleet of wind plants within hundreds of miles of each other to show large variations.[4] In a jurisdiction with greater wind capacity, variations across the fleet can be significant, with quite possibly the same effect as more than one 1,000 megawatt-sized unit of conventional generation. As this can happen relatively frequently to this extent compared to conventional plant failures, the risk is that this could easily overwhelm the “normal” synchronized reserve and render it unavailable in the event of a conventional plant failure, which is the purpose for which it was designed.

Komanoff’s following conclusion (albeit qualified to the PJM experience) is very questionable.

“We thus have, for the PJM system, the answer to the central question of wind power’s displacement of fossil fuels…88%-96% of the “theoretical” fossil fuel savings for wind power — the savings that would be calculated from equating each kilowatt-hour of wind generated with a kilowatt-hour of fossil fuels avoided — remain after allowing for reserve requirements.”

Again all this relates to the PJM system, which is not illustrative of the impact of any consequential amount of wind penetration. Finally and somewhat understandably considering the minimal amount of wind penetration involved, no consideration is made of the necessarily increased inefficient operation of the reserves due to wind, and the effects of this on fuel consumption and CO2 emissions in systems with greater wind penetration.

Komanoff goes on to claim, “The takeaway from the PJM interview — that reserve maintenance uses up only a small percentage of the fossil fuel savings from wind power generation — is evidenced in essentially every major study of wind integration on utility grids.” If Komanoff is relying on the UK paper (Gross et al) for this, which he cites, it should be noted that a large number of the studies reviewed in the report predate the turn of the century. Further, one of the studies included, from the German Energy Agency (dena), dated 2005, does not extend results past about 17 per cent penetration, whereas many of the others project to 30-40 per cent, which is totally unrealistic based on practical experience in Denmark and Germany. The author of the UK report (Gross) labels Germany’s dena low results as “outliers”. This is interesting considering that Germany is the most experienced and its report one of the most recent of those reviewed by Gross. The use of the term “outlier” may be significant if all the observations (theoretical projections in the majority of cases cited) relate to a natural phenomenon or underlying process with a central tendency, which they do not. They are individual theoretical projections based on models.

Continuing on from the Komanoff quote in the previous paragraph, he cites two reports he selected “at random” in support of his claim, but both are from low penetration jurisdictions, the state of Minnesota and the UK (Gross), the latter of which can be questioned, as already indicated. As a simple example, Gross appears to attempt to further discredit the dena study results, which is not favourable to wind, by referring to two other “German” studies, whose results lie within the more favourable (and more populated) range. They do not originate from German sources, but are Austrian (2004) and European Union (1992) studies. A more detailed analysis of the Gross report follows in Part II.

In summary, although normal operating reserves may be adequate in very small wind penetrations, the situation changes significantly when penetrations exceed a few per cent in MWh terms. The requirement for additional backup of a different, and more extensive, nature solely for wind becomes apparent.

Load Following Capabilities of Fossil Fuel and Nuclear Plants

With respect to coal plants, Komanoff recalls the rule of thumb that “…to accommodate diurnal variations [between day and night] in aggregate demand, large, modern coal-fired units could be banked down from 100% load to as little as 25% of full capacity, and then back up again, without incurring “thermal stress” that could lead to tube leaks, pipe cracking, or other damage, so long as the transition was gradual rather than abrupt.” (emphasis added) He goes on to suggest “If this rule of thumb still holds…” (emphasis added), that coal plants could follow wind’s continuous volatile output throughout the day. Let alone the very questionable major jump in logic from one “cycle” per day to many, the focus is on damage to the coal plants’ components (which is important enough), but no mention is made of fossil fuel consumption and CO2 emissions resulting from this abnormal and inefficient mode of operation, which are at least equally important considerations.

With respect to nuclear plants, Komanoff claims that “Most nuclear power plants also have considerable load-following capability…” (emphasis added), implying that they would be able to respond to wind’s continuously volatile output. This is a very questionable assertion. France, which has substantial nuclear plant capacity, is cited as an example of the need for nuclear plants to respond to changes in demand and this is suggested to illustrate their ability to mirror wind’s volatility, but this has limitations. Further, France has less than 1 per cent wind penetration and also has gas and hydro generation capacity, both of which are more likely candidates to be used to respond to demand and wind fluctuations throughout the day.

Part II completes the case study, focusing on the importance of capacity value, and includes some analysis of the Gross et al paper, which provides additional case study material.

---

[1] Sharman, Hugh, Incoteco (Denmark) ApS, Planning for Intermittency: The Importance of Evidencefrom Germany and Denmark, slide 24, UK ERC Workshop, Imperial College, July 2005 (emphasis is Sharman’s) http://www.ukerc.ac.uk/Downloads/PDF/05/050705TPASharmanpres.pdf

[2] E.ON Netz, Wind Report 2005, page 9. http://www.eon-netz.com/Ressources/downloads/EON_Netz_Windreport2005_eng.pdf

[3] Hoppe-Klipper, Martin, System studies and best practices – Germany, Managing Director, deENet, Energie mit System (a consortium of 90 research institutions and service providers in Germany) http://www.deenet.org/ . Presentation (see slide 13) was made at the Large Scale Integration of Wind Energy EWEA Policy Conference in 2006, and can be accessed through the European Wind Energy Association site at http://www.ewea.org/fileadmin/ewea_documents/documents/events/2006_grid/Martin_Hoppe.pdf. I caution the reader to be aware that the CO2 emissions reductions and additional costs (based on the dena study) in the deENet presentation have to be examined carefully and understood before drawing conclusions and making associated claims.

[4] Adams, Tom, Transforming Ontario’s Electricity Paradigm: Lessons Arising from Wind Power Integration, May 2009, keynote address for the Annual General Meeting of the Professional Engineers of Ontario. http://tomadamsenergy.com/site/uploads/2009/05/keynote-for-peo-may-2009-transforming-ontario_s-power-system.pdf. See also Hugh Sharman note 1 above.