The cost and value of variability in electricity generation

The Bloomberg terminal offers an LCOE function provided by its New Energy Finance unit. The function calculates the levelized cost of electricity for a number of generation technologies. The LCOE is the discounted lifetime cost of a generating one unit of electricity from a particular plant type taking into account all capital and operating costs. Shown here are the results for solar thermal, offshore win, solar PV, biomass and municipal waste incineration, geothermal, wind onshore, coal fired, natural gas combined cycle, and landfill gas:

But just because you can crunch the formula doesn’t mean the results are meaningful. Average cost is interesting, but it ignores two things that are critical to properly evaluating different generation technologies.

The first is value. Not all electrons are the same. Where power is traded in competitive wholesale markets, the price varies dramatically across hours of the day, seasons of the year and system conditions. Building the cheapest type of generation may be a good example of being penny wise and pound foolish if that technology produces power when it is least valuable.

Here’s a graph of the output of wind generators in ERCOT, the Texas system during one week in August, overlaid on top of a graph of the system load. The negative correlation is stark.

My colleague, Paul Joskow, has a much more detailed exposition of the confusion between cost and value in this paper. I took the graph above from his presentation on the issue at our Center’s semi-annual Energy and Environmental Policy Workshop last week.

The second issue is system cost. Both solar and wind generation can experience sharp volatility in output in short intervals of time. Below is a graph from another paper presented at the Workshop by Stan Reynolds of the University of Arizona. It shows the solar output from PV test facility near the Tucson airport during an August day in 2008.

The sharp fluctuations result from the traveling cloud cover. In order to maintain a stable electricity system in the face of this highly variable supply, the system operator must take appropriate precautions. In particular, the operator must provide additional reserve generation.

The need to account for system costs is widely acknowledged. But LCOE calculations seldom take this into account. And determining the correct value is a difficult task. There is a large literature on the problem, although most of it fails to properly grasp the problem, and estimates lack much empirical foundation. In their paper, Reynolds and his colleagues move the ball forward with a well constructed model that properly recognizes the economic forces at hand and that can be calibrated to the specific data for a given region and technology. In their particular analysis, they find that these extra system costs add only a small amount to the effective cost of the solar generation. Of course, their paper, too, only captures some of the consequences of intermittency. But it’s a good contribution towards a proper accounting of the full costs.


  1. Posted November 10, 2011 at 7:11 am | Permalink

    Including a capacity factor doesn’t resolve the problem of valuing the variability. Regardless of the average number of hours a generator produces, we are still left needing to determine (i) whether it produces in high value hours or low value hours, and (ii) the system costs created by variability around the mean.

  2. Posted November 10, 2011 at 6:07 am | Permalink

    Very interesting post. They have tax rates as an input, which in itself is highly variable too from country to country with clean-energy incentives.

    I notice that the DOE does include a capacity factor in their calculation.

Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: