Monday, January 25, 2010

Can interconnected windfarms replace baseload power plants? Part I, by Bill Hannahan

My apologies to Bill for posting the wrong version of his paper yesterday. When I down loaded it yesterday, for some reason an older draft version popped up. No doubt this was the consequence working while half asleep. At any rate here is my amended post, with Bill's latest draft of his paper.

Bill Hannahan is a retired engineer. Bill is a frequent commenter on Internet posts regarding Energy related issues. Although he has no blog of his own, Bill is highly regarded by Nuclear Bloggers as someone who if both intelligent, well informed and articulate. Some time ago Bill Bill wrote me to inform me of a paper he had written on the often cited paper by Cristina L. Archer and Mark Z. Jacobson, Supplying Baseload Power and Reducing Transmission Requirements by Interconnecting Wind Farms. Bill had attempted to get his critique of the Archer-Jacobson paper on The Journal of Applied Meteorology & Climatology (JAMC), but had been frustrated by the refereeing process. After Bill submitted his paper, and it was sent to Mark Jacobson for comment. Bill responded with what he regarded as a much improved paper that responded to Jacobson's comments. When Bill's second round paper was submitted to Jacobson for comments, Jacobson refused any further cooperation. The editor of he JAMC then offered to publish Bill's original paper. Bill argued that the revised paper should be published, and at that point they were stuck. So bill withdrew his paper, and was at the time he wrote me interested in posting it on the Oil Drum, together with an account of his attempt to have it published on the JAMC. I offered to Bill that if his Oil Drum effort was not sucessful, I would be most happy to post his paper and the account oh his attempt to have it published on Nuclear Green

Can interconnected windfarms replace baseload power plants, Part I

The Journal of Applied Meteorology & Climatology (JAMC) published a peer reviewed paper by Stanford professor Mark Jacobson and Cristina Archer called

“Supplying Baseload Power and Reducing Transmission Requirements by Interconnecting Wind Farms”

A review comment on the electrical engineering portion of the analysis was submitted in accordance with the published procedure which calls for two rounds of comment/author response, with publication of the second round. The authors submitted a response to the first round comment. The following final comment was submitted.

A review Comment


Supplying Baseload Power and Reducing Transmission Requirements by

Interconnecting Wind Farms


Department of Civil and Environmental Engineering, Stanford University



The paper makes this claim;

“It was found that an average of 33% and a maximum of 47% of yearly averaged wind power from interconnected farms can be used as reliable, baseload electric power.”

This claim is not supported by the analysis for the following reasons;

1 On page 14 the authors wrote, “The area of interest was within the Midwestern United States, previously identified as one of the best locations for wind power harnessing over land.”

The location chosen for the distributed wind model is the best in the country, as indicated by their average calculated capacity factor of 0.45, about a third higher than the national average, also partially explained by the large machines assumed. Wind power produces about 1% of our electricity, and most existing wind farms are on prime locations. If wind is to be a major source of energy, most arrays will be on locations where conditions are not as good. The study should have used an average area, for its conclusion to be applicable to our energy problem.

2 The parameters for power plant reliability are set down in IEEE Std. 762-2006, “IEEE Standard Definitions for Use in Reporting Electric Generating Unit Reliability, Availability, and Productivity.” Here are the relevant excerpts.

8.13 Net capacity factor (NCF): Equals 100 times net actual generation / the energy that could have been produced by a unit in a given period of time if operated continuously at maximum capacity.

3.22 Planned Outage Factor (POF): The fraction of a given operating period in which a generating unit is not available due to planned outages. Planned outage: The planned outage state is where a unit is unavailable due to inspection, testing, nuclear refueling, or overhaul. A planned outage is scheduled well in advance.

3.26 Unplanned Outage Factor (UOF): The fraction of period a generating unit is not available due to unplanned outages. Unplanned outage: The unplanned outage state is where a unit is unavailable, but is not in the planned outage state. (Unplanned outages are subdivided into maintenance outages and forced outages) Maintenance Outage: A maintenance outage can be deferred beyond the end of the next weekend, but requires that a unit be removed from the available state or another unplanned outage state before the next planned outage.

6.11 Maintenance outage hours (MOH) The phrase maintenance outage hours represents the number of hours a unit was in a maintenance outage state.

3.14 Forced Outage Factor (FOF): The fraction of a given operating period in which a generating unit is not available due to forced outages.

6.10 Forced outage hours (FOH) The phrase forced outage hours represents the number of hours a unit was in a Class 0, Class 1, Class 2, or Class 3 unplanned outage state.

A Class 0 unplanned outage results from the unsuccessful attempt to place the unit in service.

A Class 1 unplanned outage requires immediate removal from the existing state.

A Class 2 unplanned outage does not require immediate removal from the in-service state, but requires removal within 6 h.

A Class 3 unplanned outage can be postponed beyond 6 h, but requires that a unit be removed from the inservice state before the end of the next weekend.

3 From the paper;

“The intermittency of wind is directly transmitted into wind power, which dramatically reduces the marketing value of wind (Milligan and Porter 2005). On the other hand, because coal combustion can be controlled, coal energy is not considered intermittent and is often used as “baseload” energy. Nevertheless, because coal plants were shut down for scheduled maintenance 6.5% of the year and unscheduled maintenance or forced outage for another 6% of the year on average in the United States from 2000 to 2004, coal energy from a given plant is guaranteed only 87.5% of the year, with a typical range of 79%-92%.... "Firm capacity" is the fraction of installed wind capacity that is online at the same probability as that of a coal-fired power plant.”

The authors have redefined reliability. First they use the term, "Firm capacity," which is not found in IEEE-762, and then when they write their conclusion it is “reliable”. A two week wind lull in the middle of a 100 year heat wave, affecting several gigawatts of wind farms over a wide area, is no more problematic to the authors than a routine two week maintenance outage of one 0.3 gigawatt conventional power plant, scheduled a year in advance for the off peak season.

Reliability and capacity factor are two separate and distinct parameters. This report blurs the distinction. During a scheduled outage the power output is guaranteed to be zero, there is no uncertainty. Maintenance and refueling outages scheduled long in advance reduce a plants capacity factor, not its reliability. Utilities plan for these events just as they plan for the fact that photovoltaic cells do not produce electricity at night. The report compares the scheduled down time of conventional power plants with the unscheduled unpredictable downtime of wind power. This is an apples to oranges comparison that does not prove that wind power can replace conventional power plants. Actually it proves wind power is not as reliable as conventional power plants.

4 Maintenance outages can be delayed until after the following weekend, 48 hours or more. This allows grid managers to arrange for replacement power before the problematic plant is shutdown, without cutting into their spinning reserve capacity. They can prepare an unused plant in cold shutdown for operation, they can reschedule a planned outage for another plant to a later date, they can accelerate the completion of in progress maintenance on another plant, they can arrange for transmission line capacity to bring in power from a distant location. They can also gather the equipment, tools and spare parts to make the outage work efficient and quick.

Wind lulls cannot be delayed, scheduled or rescheduled. Reductions in generation due to unforcast wind lulls must be replaced in real time using up spinning reserve capacity. Since maintenance outages can be delayed 48 hours or more, they are not dependent on spinning reserve and are not equivalent to the unscheduled, unpredicted loss of wind power. Therefore maintenance outages should not be included in the comparison with wind intermittency.

The North American Electric Reliability Council database for coal generation in the years 2000-2004 show that the average forced outage factor for that period was 3.96%. Applying this to the paper’s figure 3, for 19 interconnected wind farms, shows that only about 10% of data plate rated wind power would be available with the equivalent FOF on an annual basis.

5 The Forced Outage Factor for natural gas plants during the same period, 2000-2004, was 2.73%. The fraction of rated wind power available with the equivalent FOF of natural gas is approximately 3%.

6 Hydro power achieved a FOF of 2.5%. The fraction of rated wind power available with the equivalent FOF of hydro power is approximately 2.5%.

7 Nuclear power achieved a FOF of 2.52% between 2003 and 2007. Nuclear power plants achieved a remarkably low forced outage rate of 1.64 outages per plant per year. The average duration of a forced outage was 135.6 hours per outage. When a forced outage occurs, grid managers soon have an estimate of how long the outage will last and they arrange for alternate generating capacity. They do not rely on spinning reserve to cover the loss for the entire outage. It effectively becomes a scheduled outage. Deleting all but the first 24 hours of each forced outage to compare with wind, gives 39 hours of wind equivalent forced outage per plant per year. The wind equivalent FOF is 0.45%. The fraction of rated wind power available with the equivalent FOF of nuclear power is essentially zero.

8 Wind conditions cannot be reliably forecast more than 24 hours in advance, therefore reliability comparisons with wind power should only be based on the first 24 hours after a problem is detected. Some Class 3 outages can be delayed more than 24 hours. Those hours should be subtracted from the total forced outage hours for purposes of comparing with wind reliability.

9 Had the paper compared wind power with a representative mix of conventional power plants supplying the U.S. grid, the results would have been much less favorable to windpower.

10 The U.S. 48 state power supply is divided into three largely independent grids, the eastern grid, the western grid and the ERCOT grid which covers most of Texas. The ERCOT grid has approximately the same area as the test area in this report, 722,000 sq. km, in fact they overlap substantially. ERCOT has 8,000 MW of wind capacity now with plans to nearly double that in the future.

The average rating for a fossil fueled power plant is 300 MW. The grid always has enough spinning reserve to cover the failure of one of these plants, however, if there is an unexpected drop in the wind over a large area containing many wind farms, it is like losing several conventional power plants at the same time. A recent unforcast wind lull in Texas gave us a preview of this.

“The grid operator went directly to the second stage of an emergency plan … System operators curtailed power to interruptible customers to shave 1,100 megawatts of demand within 10 minutes… ERCOT said the grid's frequency dropped suddenly when wind production fell from more than 1,700 megawatts, before the event, to 300 MW when the emergency was declared”.

If the wind farms had been fully built out that same lull would have produce a much larger loss of generation.

Treating a large array of windfarms as if they are a single fossil plant, or even as an array of independent fossil plants, is not realistic. The possibility of common mode failure due to widespread meteorological conditions resulting in a large drop in generation dramatically increases the spinning reserve required to assure grid reliability. The cost, fuel consumption and emissions associated with maintaining a larger spinning reserve for wind arrays should be attributed to the windmills.

The sudden wind lull on the ERCOT grid was not forecast even a few hours before it happened. Giving wind power credit for a continuous reliable forecast 24 hours in advance, as was done in previous points, is extremely generous.

11 From the paper;

“Figure 3 shows that, while the guaranteed power generated by a single wind farm for 92% of the hours of the year was 0 kW, the power guaranteed by 7 and 19 interconnected farms was 60 and 171 kW, giving firm capacities of 0.04 and 0.11, respectively. Furthermore, 19 interconnected wind farms guaranteed 222 kW of power (firm capacity of 0.15) for 87.5% of the year, the same percent of the year that an average coal plant in the United States guarantees power.”

The 12.5% of unscheduled wind power below 15% amounts to 45.6 days per year. They are most likely to occur during the worst of times, extreme heat wave and cold snap conditions. This is not an acceptable level of reliability for multi GW arrays of power plants subject to common mode failure, especially when they are most likely to fail under extreme conditions.

12 The paper compares the reliability of one coal plant with the reliability of an array of 19 wind farms. To make it an apples to apples comparison, the same analytic technique should be applied to an array of 19 conventional plants, which is more like a real grid, and would have much higher reliability.

13 The word reliability has been stretched so badly that it is almost meaningless. A better characterization is “controllability, stability and predictability”. Consider a hydroelectric reservoir with the same data plate rating as the proposed wind array, and enough water inflow to run the generators at the same capacity factor, 45%. A grid manager heading into a summer heat wave would much rather have the hydro plant than the wind array, because the hydro plant can ramp up to 100% for a few hours during peak demand, and back to 30% the rest of the day. Both examples have the same “firm capacity”, but hydro has controllability, stability and predictability, the key factors grid managers need to maintain proper voltage and frequency on the grid. Reliable controllable, stable, predictable kWh's are worth much more than unreliable, uncontrollable, unstable, unpredictable kWh's.

14 We can have an all fossil powered grid, an all nuclear powered grid, or an all hydro powered grid, but we cannot have an all wind and/or solar powered grid, even with large arrays of interconnected machines. For that we must add an enormous amount of expensive energy consuming storage capacity or a large amount of conventional generation capacity to provide the required controllability, stability and predictability.

Coal plants are designed for a 40 year lifespan, and the average age of U.S. coal plants was 40.7 years in 2007. If we go with a massive wind construction program, the cost estimates should include the new backup plants that will have to be built, their fuel supply, and the new transmission lines and energy storage systems built to support windpower, but promises of cheap wind power do not include these costs.

15 Intermittent sources like wind and solar can contribute energy to the grid but they do not contribute stability or reliability, in fact they suck up the stability provided by conventional power plants, reducing grid stability and reliability. For this reason it is misleading to say that wind and solar can replace any fraction of baseload power plants.

“Traditional power stations with capacities equal to 90% of the installed wind power capacity must be permanently online in order to guarantee power supply at all times."

The only savings from wind and solar is the fuel not burned when the wind blows or the sun shines. To avoid raising anyone’s electric bill the cost of wind and solar kilowatt hours must not be greater than the cost of the fuel saved, 5.7 cents per kWh for natural gas, 2.4 cents per kWh for coal and 0.5 cents per kWh for reactor fuel assemblies.

16 Because wind power is often best when demand is low, windfarms sometimes find themselves paying the grid to take their power so that they can collect the wind generation subsidy payments.

“In the first half of 2008, prices were below zero nearly 20 percent of the time. During March, when negative prices were most frequent, prices were below zero about 33 percent of the time.” (Knowledge Problem, 2008)

17 For the paper’s conclusion to be useful the recommendation must be affordable.

From the paper;

“A final benefit of interconnecting wind farms is that it can allow long-distance transmission from a common point, where several farms are connected, to a highload area to be reduced with little loss of transmitted power. Suppose we want to bring power from N independent farms (each with a maximum capacity of, say, 1500 kW), from the Midwest to California. Each farm would need a short transmission line of 1500 kW brought to a common point in the Midwest. Between the common point and California, the size of the transmission line would normally need to be N x 1500 kW. However, because geographically disperse farms cause slow winds in some locations to cancel fast winds in others, the long-distance transmission line could be reduced by 20% (to N x 1200 kW) with only a small loss (2% with N x 19) in overall delivered power (Fig. 3).”

Assume California built a 1,500 MW nuclear power plant at Amarillo Texas. A 1,500 MW transmission line to California would be at full capacity 90% of the time and zero capacity 10% of the time during refueling and maintenance outages. At $4.00 / data plate Watt, the nuclear plant would cost $6 billion, and the transmission line to California would cost about $1.47 per kW mile, $2.5 billion (land not included), total cost $8.5 billion plus transmission line land cost. Or, simply build the nuclear plant in California and save the transmission line cost and line loss.

To replace the 1,500 MW nuclear plant with "reliable" wind power we need to install 19 wind farms, each with a data plate rating of 474 MW, 9,000 MW total, (1,500 MW x 0.9 capacity factor / 0.15 reliability factor). The latest cost estimate for T Boon Pickens 4,000 MW wind project is $12 billion. Subtracting $2 billion for transmission lines, the cost of the installed windmills is $2.50 per data plate Watt. The windmills for the California project will cost $22.5 billion, and they will need to be replaced in 25 and 50 years, whereas the nuclear plant will last 60 years.

To connect these farms to a central point will require on average 100 miles of 474 MW transmission line each. At $1.47 per kW mile, the collection grid will cost $1.3 billion, plus land cost. The 9,000 MW wind array could limit its transmission line capacity to 7,200 MW in exchange for a 2% loss. We would have to build 5 transmission lines to California like the one for the nuclear plant, at a cost of $12 billion. The total cost to replace one nuclear plant with "reliable" wind power is $35.8 billion plus transmission land cost. The average capacity factor of the wind transmission lines is much lower than that for nuclear, and the average distance wind kWh’s travels is much greater than the average for nuclear kWh’s, so transmission cost per kWh is much higher for wind than for conventional plants. Clearly wind is not a practical substitute for baseload power plants.

18 Denmark has been pushing wind very hard since 1979. It has ideal conditions for wind, and produces the most wind power per capita of any country, about 150 watts per person, which is less than 10% of the U.S. consumption rate, 1,550 watts per person. Denmark has the most expensive electricity in the world, 38 cents per kWh, and generates most of its electricity with fossil fuel. When wind conditions are good, Denmark sells excess wind power to their neighbors at fire sale prices, and when wind is poor, they buy nuclear, fossil and hydro power from them. Only half of Denmark’s wind power is consumed in Denmark, about 75 watts per person, the other half is exported when wind production exceeds their capacity to consume it. If the U.S. scaled up to Denmark’s level we would have to shutdown a large fraction of our windmills on windy days because our neighbors cannot absorb so much excess power, nor can they supply that much power on calm days.

Scientific conclusions do not have to be affordable, but it is misleading to present very expensive solutions to the public without a cost estimate, as if they are practical.

19 One of the biggest drawbacks of wind power is the daily and seasonal variation in power output. Wind power is very low in the summer when electrical demand peaks, and high during spring and fall when demand is lowest. Over the entire U.S., wind power dropped 20% below average during July and August of 2006, while Electricity consumption for the nation jumped 20% above average. Nuclear power plants normally run at 100% year round due to their low fuel cost, yet production was 10% above average during July and August because they schedule refueling and maintenance outages for spring and fall when demand is low. Windfarms often schedule maintenance in the summer when wind is low to minimize production losses.

In California, windmill output at the time of peak demand during the heat wave of 2006, dropped below 4% of data plate rating for seven days. Counting on wind power for baseload capacity during a heat wave could be deadly.

This paper does not highlight these problems and the calculations homogenize a year of data in such a way that the graphs completely hide seasonal variation.

20 There are 3 billion people around the world who want to join the middle class. If the U.S. could reduce its emissions to zero instantly, the savings would soon be gobbled up by the developing world. Second and third world countries do not have large robust grids with massive baseload plant capacity sufficient to provide free backup and power conditioning for a large windmill buildup. New backup plants and fuel supply systems would have to be built along with the windmills, making wind cost prohibitive in those countries.


Clearly the real world requirements of grid management and the real world requirement for affordability and the analysis in this report do not support this conclusion;

“It was found that an average of 33% and a maximum of 47% of yearly averaged wind power from interconnected farms can be used as reliable, baseload electric power.”

A non technical person with no knowledge of the utility industry, reading this paper, will receive an inaccurate view of winds capacity to replace conventional power plants. This paper is being used by wind enthusiasts to convince ordinary citizens and political leaders that we can replace conventional baseload power plants with wind power plants, with no increase in cost or risk (Roberts 2007), (Lipow 2007), (Treehugger 2007).

Consider this quote from Popular Mechanics magazine,

“Cristina Archer, a professor of civil and environmental engineering at Stanford University, has proposed a different approach... Archer … determined that if 19 of those farms were networked together, grid operators could safely rely on more than a third of the total power generated—it would become as dependable as electricity from a traditional power plant.”

We are at a critical point in human history, making the transition from fossil fuel to other sources of energy. Getting energy right or wrong will mean the difference between life and death for hundreds of millions of people in this century. Windmills are not a practical replacement for any fraction of baseload power plants.

This ends the comment.

I would like to add one more example that came to mind after the comment was submitted that clearly illustrates the fatal flaw in this analysis.

Imagine that someone invents a cold fusion generator that is very cheap to mass produce, but can only be operated at rated power for 1 hour in every 10 hours. The “Firm Capacity” of these units is only 10%, less than one fourth the 45% Jacobson claims as the “Firm Capacity” of the windfarms.

A utility would buy enough units to cover its highest demand day, and have a totally reliable grid year round with no backup plants, no storage and no additional transmission lines required. That is because each unit can be scheduled well in advance and dispatched as needed. These are the qualities that give an energy source reliability, and they are not present with wind farms, even if they are interconnected. “Firm capacity” is not an indication of, or correlated with, or equivalent to reliability, except at the end points of 0% and 100%.

Jacobson’s entire paper hangs on the assumption that “firm capacity” and “reliability” are equivalent and interchangeable, in direct contradiction of this example, the examples in the comment, common sense and industry standards.


Robert Hargraves said...

If Hannahan were to rewrite the Jacobson/Archer paper, substituting corrections he has mentioned, what would be the result?

It would seem to be to be a simple MS engineering thesis to do a simulation of wind power production using historical wind data. Connect them together and see what the resulting power is.

arcs_n_sparks said...

Well written rebuttal; too bad it has not been accepted for publication.

I agree with Mr. Hargraves; this should be capable of being modeled using historic wind data.

Paul Lindsey said...

When I read the Archer & Jacobsen paper, I thought the capacity factor that they used was way too high. This paper by a Spanish economist who analyzed wind power worldwide from 2003-2007, gave an average US wind power capacity factor of 25.7%.

What is the location of this Midwestern wind farm with an actual 45% capacity factor?

Charles Barton said...

Paul, Archer & Jacobson did not use typical wind locations, rather they picked out prime locations in the best settings. There is a very considerable variation in the wind capacity of different locations, and West Texas, Oklahoma, Kansas, and New Mexico have some locations where capacity factors of .40 and above have been observed.

Paul Lindsey said...


I have been very skeptical of claims of wind capacity factors over 20%. The AWEA and other wind advocates consistently use a 35% capacity factor in their publications.

Can you point me to the references that show US land-based "locations where capacity factors of .40 and above have been observed", because I can't find any. I suspect that some of these observations have been instantaneous or less-than-annual values.


John Morgan said...

The account of the review process sounds odd. For Jacobsen to refuse to complete his obligation as reviewer is simply unprofessional. Do we know in what terms Jacobsen has refused cooperation?

If Hannahan has responded to Jacobsen's initial review, and Jacobsen is refusing to exercise his opportunity to comment further, surely that amounts to a decision by Jacobsen to allow the revised version to pass into publication without further comment?

The idea that the editor is not able to publish without Jacobsen's say so also seems absurd. Surely in the case of an uncooperative reviewer the editor can exercise discretion. He should publish the revision on the basis that Jacobsen has not felt there is anything further to address in the revision. But failing that, he could assign an alternative reviewer.

This just sounds fishy. It doesn't sound like the formal review process is done with yet. I hope Bill will continue to persevere with the editor, or submit to a different journal. It is really important to get these arguments into print. Failing to do so is to abandon the scientific high ground to naive flights of academic fantasy, without challenge from engineering practice and experience.

Soylent said...

I think some of the best anti-wind PR possible would be to make a simplified but still somewhat realistic electric grid simulator game that tasks the player with designing an electric grid.

The player would essentially be given dictatorial powers over a whole continent and would be tasked with building a functional electric grid, they'd have demand profiles for different areas, wind maps etc. Then they'd plopp down the infrastructure in it's built state(don't worry about the transition to this grid, just assume it as a starting state) and press "play". Past wind and solar data to cover some years would then be used to calculate the reliability over some years, toghether with monte carlo analysis of unplanned outages and grid disruptions of various kinds and some reductionist model of how a grid operator would behave and how much spinning reserve they would have.

Then they'd be given a score card that contains the cost, the CO2 intensity in grams per kWh supplied(surplus kWh, e.g. wind nobody wants to take delivery of, are simply discarded) and a map containing the extent of brown-outs/rolling black-outs and the extent of black-outs. They'd be given the option to tweak their grid and try again.

Unfortunately it's a huge pain in the arse to build such a grid and obtain all the relevant data for some time-period.

DW said...

Here is what the are raving about now:


Blog Archive

Some neat videos

Nuclear Advocacy Webring
Ring Owner: Nuclear is Our Future Site: Nuclear is Our Future
Free Site Ring from Bravenet Free Site Ring from Bravenet Free Site Ring from Bravenet Free Site Ring from Bravenet Free Site Ring from Bravenet
Get Your Free Web Ring
Dr. Joe Bonometti speaking on thorium/LFTR technology at Georgia Tech David LeBlanc on LFTR/MSR technology Robert Hargraves on AIM High