Changing a reputation: Wind energy projections are getting more accurate

A few posts ago, DNV GL’s Steve Jones shared some reflections on WINDPOWER 2013, in which he pointed to many signs of the maturing of the wind industry. Today, I am reflecting on another of those signs—evidence of improvement in the industry’s ability to accurately predict the performance of wind power projects before they are built. This improvement, assuming it continues, should help reduce the cost of capital to wind projects and/or increase the amount of debt they can support. Both of these impacts will help to reduce the cost of wind energy and increase its competitiveness in the energy markets.

The wind industry has a reputation for overestimating the power that will be produced by wind energy projects and, to some, DNV GL has a reputation for being “too conservative” in its energy assessments. The industry’s reputation is a well-deserved. Data presented by DNV GL and other organizations clearly shows that projects built between 2000 and 2009 perform approximately 12 percent below expected levels when the energy assessments were completed by organizations other than DNV GL. DNV GL’s reputation for being overly conservative is, in fact, not supported by data.  The data show that projects for which we have predicted energy production between 2000 and 2009, on average perform 3 percent below their predicted levels. The reasons for the differences are rooted in differences in DNV GL’s unique methodologies and assumptions, more than any specific tools or knowledge within the company.

Figure 1: Project Average Results for 2001–2009
Graph 1

DNV GL has been chronicling the performance of the industry since 2006 when documented wind project actual performance relative to pre-construction energy estimates. Since then, we have updated this work and encouraged open dialog throughout the industry, specifically what can be done to address the issue. The investor community has also been pressuring the industry to improve its energy estimating practices. In response to these pressures, the industry has been conducting additional research and adjusting energy assessment methods.

An important question is how well the adjustments the industry has made have reduced the historical over prediction bias. DNV GL sought to answer this question in the Project Performance White Paper published in May 2013. In this edition of the white paper, DNV GL segregated the analysis into two periods, 2000–2009 and 2010–2012. The results of the analysis show that the over predication bias on projects predicted by others in the industry has reduced from approximately 12 percent  to between 2 and 4 percent—depending on how projects with suspected curtailment are treated. The data show a DNV GL bias of less than  1 percent. In my opinion, the change in the DNV GL results between the 2001–2009 data and the 2010–2012 data is not statistically significant. The data set is relatively small and I am not aware of shifts in our methods that should have changed the bias in the relevant time period. However, changes made in our methods in 2011 and 2012 should be evident in future results. On the other hand, I am aware of changes in energy loss assumptions and other factors that can be expected to reduce the bias in the energy assessments completed by others. It is not clear that these changes would be expected to reduce the bias as much as has been reported, however with additional data, the value of the remaining bias will be better defined.

Figure 2: Project Average Results for 2010–2012
Graph 2

The prior point regarding curtailment is interesting. As errors in estimating the wind resource decrease, and wind energy penetrations on the grid are increasing, curtailments of wind power plants to either relieve congestion on the grid or in response to an oversupply of energy on the grid, are increasing. It appears that more time and attention needs to be given to estimating energy losses due to curtailment. There are significant challenges with accurately estimating future curtailment due to the uncertainties as to what will be built in the future, and where it will be built, but developing better strategies for estimating curtailment could be the next important step in increasing the financial community’s confidence in wind project energy assessments and overcoming the industry’s over prediction reputation. In the meanwhile, I encourage anyone with operating project data to contact me at Robert.Poore@dnvgl.com so that we can include this data in the data base and continue to provide facts which investors and the rest of the industry can use to evaluate the reliability of wind project energy estimates.

By: Robert Poore, senior advisor, Renewable Energy Services, DNV GL

Leave a comment

Leave a Reply

Your email address will not be published. Required fields are marked *