Brookings: Wind and Solar Technology Fail

By Kent Hawkins -- May 29, 2014

“Even with carbon emissions valued at $50 per metric ton, nuclear, hydro and natural gas combined cycle generation plants have far more net benefits than either wind or solar.”

The recent paper by Charles Frank of the Brookings Institution, “The Net Benefits of Low and No-Carbon Electricity Technologies” provides a reasonably broad, detailed analysis of the lack of value in pursing policies of implementing wind and solar industrial-scale generation plants to reduce carbon emissions. This analysis, however, while on track, misses some very important considerations that strengthen the already negative verdict.

In summary, the paper finds:

• Even with carbon emissions valued at $50 per metric ton, nuclear, hydro and natural gas combined cycle (combined cycle gas turbine, or CCGT) generation plants have far more net benefits than either wind or solar, because the latter have a very high capacity cost per megawatt (MW), very low capacity factors, and low reliability.
• In the absence of carbon pricing new “no-carbon” plants will tend to displace low-carbon gas CCGT rather than high-carbon plants.
• Direct regulation of CO2 emissions for coal-fired plants, as proposed by the EPA can be as effective as carbon pricing.

The rest of the post below extends the analysis.

Full Costs for Wind and Solar

A significant issue is taking the cycling and balancing costs to be “quite modest”, based on reports by Ellerman and Marcantonini, and Van Bergh et al. These are dependent on questionable references themselves, for example Gross et al and Milligan. I have previously critiqued the Gross et al paper, which appears impressive, but on closer examination is very questionable, and Milligan in a general review of wind is also questionable. Interestingly, a 2013 presentation by Marcantonini simply assumes a 2€ per MWh for additional balancing due to wind. From these Frank concludes “the variability renewable energy production adds little to the variability of residual demand….”

This is simply not the case. Wind and solar production are highly variable in the short term like fluctuations in load. The combination of these events in real time is in effect a net load that must be dealt with by reliable generation plants. The combination of wind, solar and load has a greater range in variation than each alone, in spite of assertions by the American Wind Energy Association (AWEA) that this is calculated by taking the square root of the sum of the squares of these events in real time.

A review of the published five-minute production data by the Bonneville Power Administration (BPA) shows that the real time (not statistical expectation over time) combined effect of wind and load (net load) increases the total range in variation over load alone by 45% for January to June 2011, and by 51% for the same period in 2010.

The effect of the additional ramping between five minute intervals of the net load versus load alone and caused by wind is as follows:

• For the period January to June 2011, there were increased ramping events up to over 1,000 MW and decreased ramping up to 190 MW.
• Using a smaller sample of January 20-26, which was chosen by a wind proponent in support of the “square root of the sum of the squares” approach, the distribution of increased ramping over load alone was 56% increased, 40% decreased and 4% unchanged.
• In the smaller sample, the change in ramping was up to 261 MW increased with up to 98 MW decreased. The statistically expected number of outliers in this period outside of three standard deviations would be about 5. The actual experience was 36, or about one every four hours on average. In simple terms this explains why a statistical approach is not realistic for real time activities.
• A statistical measure of variation, the standard distribution, for the January to June sample was only 37 MW, and for January 20-26, 26 MW, again illustrating the inadequacy of statistical approaches.

Frank uses statistical measures to determine the impact of wind and solar, based on the variance over a 10-year period, which is derived from the square of the amount of the actual value from the mean of all values in the set. He does admit that taking shorter time intervals should increase the variance, but this misses the point that it is the actual value at any point in time that is important, because at any point in time this variance does not reflect the impact of  wind and solar.

For a number of very short periods the variance would likely be notably different. Frank’s approach over very shorter time periods, and even capacity credit to be discussed below, may have value in general capacity planning but not in assessing actual wind and solar impact on real time system operation.

Capacity Credit

In determining the contribution that wind and solar make, especially at peak load, capacity credit is the proper measure. Eon Netz in Germany is among the most experienced with high wind penetration and it has determined that the capacity credit for wind quickly falls to below 10% as penetration increases. The Eon Netz finding is supported in a  European Wind Energy Association paper (EWEA) on pages 122-124. This level is determined based on 99% system reliability. Higher capacity credits are the result of assuming 91% system reliability, which is not acceptable. Arguably even 99% is still too low.

Frank does not base his analysis on this level of capacity credit but on a higher value to determine avoided capacity (page 9 and note 4). Although there is some confidence that wind and solar will operate at some level over a year close to capacity factor experience, the confidence in very short time periods, critical to electricity system reliability, is very low.

Replacing Generation vs. Reducing Emissions

It is important to differentiate between replacing electricity generation and reducing emissions as a result. Discounting the variability of wind and solar production as Frank does, leads to missing this important factor in assessing the impact of cycling on fuel consumed and emissions produced by fossil fuel plants used in balancing wind and solar. This goes beyond start up and shut down considerations. Determining this is not easy due to lack of real time fuel consumption data and inadequate published emissions data for electricity systems.

 Carbon Pricing

Carbon pricing might be an interesting hypothetical tool, but it is not a serious policy option because of carbon leakage when applied to production, the conventional approach, and this is recognized by experts in the European Union. The conventional view based on production is a flawed one, which typically results in carbon leakage.

The way this plays out, as it has in Europe, is the taxed jurisdiction will tend to de-industrialize because of higher costs and import less expensive products from another jurisdiction with a high carbon footprint. The overall effect of global carbon reduction (the only way to look at it) is thwarted.

In theory, changing this to a consumption tax by determining the carbon inputs by product is more relevant. Imports are handled by imposing a border tax, which is credited with any product content tax imposed by the exporting country. This allows unilateral initiatives without the carbon leakage disadvantages of the production tax. Exporting countries then have the option of implementing their own versions to capture the revenue.

One of the major problems with this is its complexity because it is a much more difficult to determine the carbon content by product than by general production process. There are approaches to address some of this, such as those proposed by Helm in his recent book The Carbon Crunch, but all considered, a carbon tax is complex, subject to considerable political and bureaucratic mismanagement and likely will produce many unintended consequences.

Criticisms by Others

One is that the paper does not account for the full life cycle of carbon emissions from natural gas production and delivery. A more complete analysis would include such considerations, but it should not be assumed that wind and solar would benefit. For example wind turbine production and implementation takes considerable steel and concrete, which are major industrial emissions producers. In Germany the wind turbine industry competes with the auto industry for the amount of steel used. The mining and transport of rare earth elements and lithium for wind and solar also represent considerable environmental impacts.

Another calls for broader cost/benefit analysis than the social cost of carbon might provide, suggesting relative benefits to the country or community, in terms of social well-being (I would have added term of economic well-being for a balanced sustainability analysis). This is not an unreasonable argument, except for a couple of factors, such as the very large task involved in doing this, and although there might be some changes in the actual levels, relative changes in the conclusions are unlikely.

Summary

These considerations aside, the Frank paper makes a notable contribution to the understanding of the impact of introducing industrial-scale wind and solar into an electricity system.