20% Wind By 2030?

The DOE issued a report today claiming that the US can generate 20% of its electricity with wind by 2030 without tax credits and without running into serious grid-balancing problems. While the report has received a lot of praise from the likes of Joe Romm and Ezra Klein, among others, I have to say that it left me somewhat incredulous.

What bothers me is the contention that the variability of wind really doesn't matter. According to the report:

A statistical analysis of the load net of wind indicates the amount of reserves needed to cope with the combination of wind and load variability. The reserve determination starts with the assumption that wind generation and load levels are independent variables. The resultant variability is the square root of the sum of the squares of the individual variables (rather than the arithmetic sum). This means that the system operator, who must balance the total system, needs a much smaller amount of reserves to balance the load net of wind. Higher reserves would be needed if that operator were to try to balance the output of individual wind plants, or all the wind plants aggregated together in isolation from the load.

For a concrete case, it offers:

A hypothetical example is offered to calculate reserve requirements. Say that system peak load for tomorrow is projected at 1,000 MW with a 2% forecast error, which makes the forecast error (i.e., expected variability of peak load) equal to 20 MW. Wind generation for a 200-MW wind plant in that balancing area is predicted at a peak hour output of 100 MW with an error band of 20%. The expected variability of peak wind generation, then, is 20 MW. Assuming that these are independent variables, the total error is calculated as the square root of the sum of the squares of the individual variables (which is the square root of 2 × 20 squared, or 1.41 × 20, which equals 28 MW). Adding the two variables to estimate reserve requirements would result in an incorrect value of 40 MW.

The problem is, wind doesn't really work this way. The "expected variability of peak generation" is an estimate. Even with very good forecasting techniques, it's sometimes going to be too small. In any case, you need enough dispatchable reserve to make up for shortfalls in relatively extreme situations--and in a grid with 20% wind, that means enough natural gas or hydropower to make up nearly all the variability. I'm also disturbed that the report talks a lot about hypothetical "studies" while paying very little attention to real-world experience with wind power in Europe. Far from the rosy picture painted in the report, countries like Denmark have not really gotten their money's worth out of their investment in wind power--see, for instance, this study of how Denmark balances its power grid in practice. It concluded that:

Denmark has the most intense wind carpet in the world, with a total of 3000 MW installed by the end of 2003— equivalent to 0.88 kW of wind energy per person in west Denmark. The average annual load factor for the wind turbine carpet in west Denmark is measured at approximately 20%. There are considerable and often rapid output variations throughout the day and throughout the year. Accurate forecasting of wind speeds is still difficult and output rarely matches demand, sometimes dropping below zero as stalled wind turbines still require power for their steering systems.

The variations, which are inherent in any wind energy system, can be readily accommodated in west Denmark because there are very strong electrical connections to the much larger grid systems of Norway, Sweden and Germany that can absorb these variations, particularly due to their reliance on rapid-reacting hydropower. Countries such as the UK, which operate an ‘island’ grid, will find it difficult to do this with slower-reacting thermal power stations and may thus have to limit their reliance on wind power.

If this is any indication, the DOE report vastly underestimates the challenges of integrating large amounts of wind generation into our power grid. To their credit, though, they do get this much right:

Reliability planning entails determining how much generation capacity of what type is needed to meet specified goals. Because wind is not a capacity resource, it does not require 100% backup to ensure replacement capacity when the wind is not blowing. Although 12,000 MW of wind capacity have been installed in the United States, little or no backup capacity for wind energy has been added to date. Capacity in the form of combustion turbines or combined cycle units has been added to meet system reliability requirements for serving load. Thinking in terms of “backing up” the wind is not appropriate because the wind capacity was installed to generate, low-emissions energy but not to meet load growth requirements. Wind power cannot replace the need for many “capacity resources,” which are generators and dispatchable load that are available to be used when needed to meet peak load. If wind has some capacity value for reliability planning purposes, that should be viewed as a bonus, but not a necessity. Wind is used when it is available, and system reliability planning is then conducted with knowledge of the ELCC of the wind plant. Nevertheless, in some areas of the nation where access to generation and markets that spans wide regions has not developed, the wind integration process could be more challenging.

Translation: Wind can "work" because it's not expected to substitute for baseload: i.e., even building 300GW of new wind turbines will not obviate the need to build new nuclear plants to provide carbon-friendly dispatchable baseload power.