In Ørsted's wake: when forecasts meet facts on wind yields

Despite all the work that goes into finding a suitable location for a wind farm, assessing the wind resource and designing the best layout and technology to suit site conditions and maximise power yields, actual output can still fall short of expectations. Windpower Monthly investigates the reasons.

Ideally, multiple meteorological masts would be erected to give more accurate predictiones across a site, but this would be too expensive (pic: Wind Measurement International)
Ideally, multiple meteorological masts would be erected to give more accurate predictiones across a site, but this would be too expensive (pic: Wind Measurement International)

Google Translate

At Windpower Monthly’s recent Asset Life Optimisation Forum in Amsterdam, several speakers suggested that projected energy yields from wind farms or individual turbines were often lower than expected.

Testing this hypothesis is not easy, as relatively few wind-farm operators publish their forecast yields, possibly as a result of some inaccurate projections in the early days.

However, projections and actual yields from some projects can be compared, and the results are mixed.

The Danish offshore-wind operators have a good track record, with some yields surpassing predictions and others falling short — but not by very much.

It is more difficult to forecast onshore energy yields — particularly in hilly terrain where the best wind speeds are to be found. Recent reports from France, Germany and the US suggest that the average deficiency now lies between 1% and 10%.

Estimating energy yields

Perhaps the most difficult quantity to predict is site wind speed or, more accurately, appropriate wind speeds for each point of the compass, especially in hilly terrain.

Ideally, multiple meteorological masts would be erected, but this would be expensive, so a combination of site measurements and numerical-flow modelling is usually used instead.

Site measurements need to be correlated with data from meteorological stations that have historical wind records, preferably going back at least 20 years, in order to establish appropriate long-term estimates of the sectoral wind speeds.

But despite recent improvements in numerical-flow modelling and the use of remote-sensing techniques to take measurements at proposed wind-farm sites, the accuracy of wind-speed predictions in complex terrain is probably the biggest single factor that leads to uncertainty in energy-yield projections.

Once wind-speed estimates have been formulated, wind-turbine power curves are calculated to work out the annual energy yield.

For wind farms, allowances need to be made for energy losses due to wake interactions. Other considerations to estimate losses and uncertainties to keep in mind are summarised in the table below.

The accuracy of the future long-term wind resource is very difficult to predict. Some authorities think global warming will result in increased wind speeds, while others believe the opposite — although this depends on the location.

Data published by the UK government suggest that average wind speeds declined by 2% over the period 2010-18. If the trend continues energy yields — and project returns — could be impacted.

Results from Europe and America

German wind energy institute DEWI published an analysis in 2013 of more than 2,500 energy-yield assessments (DEWI Magazin 42, 2013).

This showed that in 40% of cases the energy delivered was within +/-5% of the predicted value in France, and 24% were in that range in Germany.

But for 45% of the projects, the yields were overestimated by more than 10%, with the average overestimation at 7% in France and at 10% in Germany.

A recent analysis of data from 56 projects in the US found that yields were overestimated by 6.7% on average, although that figure came down to 5.5% for projects commissioned after 2011 (Understanding Biases in Pre-Construction Estimates, Monte Lunacek et al, 2018 J. Phys: Conf Ser).

Another report from consultants Arcvera, published this January, showed that their assessment techniques, applied to 31 wind farms with a capacity of 3,667MW, "yields an average project performance of 98.8%. As estimates of grid curtailment losses are not corrected for … the overall project performance is effectively 100%."

Analyses of wind-turbine performance were published by the California Energy Commission (CEC) in the late 1980s and early 1990s.

Data from the 1994 Wind Project Performance Summary shows the industry has advanced since that time. The report contained details of 14,577 turbines totalling 1,609MW.

Across the state as a whole, the ratio between actual and projected output was 79%.

The 166 turbines in the Pacheco resource area, in mountainous terrain south-east of San Francisco, delivered the best result — in terms of accuracy — with the ratio between actual and projected output at 104%.

The full picture is summarised in the table below. However, the industry was clearly improving its ability to predict output, as the statewide deficit in 1990 had been 31% ­— 10 percentage points below the 1994 figure.

The early days of offshore

By the time the 1994 CEC report was published, the world’s first offshore wind farm had been built and was operating in Danish waters.

Reports from the Danish Energy Agency enable assessments to be made of the accuracy of energy predictions.

Commissioned in 1991 and decommissioned in 2017, the Vindeby offshore project was built 2.5km from the coast in the fairly sheltered waters between East and West Denmark, comprising 11 Bonus 450kW turbines. The projected capacity factor was 23.9% and the actual performance is shown in the chart (below).

Although regression analysis suggests there was a steady downward trend in output, it may be noted that in some years the capacity factor exceeded the projected figure, and calculated over the first 20 years, the average capacity factor was 23.6%, which represents a 1.3% energy deficit over the period.

However, over the first 20 years’ life of the wind farm, the energy content of Danish wind was 1.3% less than the long-term average, so it could be argued that the wind farm itself delivered on its projected performance.

Although there was clearly some deterioration in performance of the Vindeby wind farm, no such trend is evident in the performance of the second Danish offshore site, at Tuno Knob.

Here, linear regression analysis suggests a slight upward trend with time, possibly due to steadily improving availability.

From the first full year after commissioning until 2018, the project performed better than expected; the average capacity factor was 31%, compared with a projected value of 30.8%.

The performance of the next four Danish offshore wind farms to be commissioned is shown in the table (below). The averages run from the first full year after commissioning until the end of 2018.

Two have delivered about 8% less energy than anticipated, one 3% less, while Samso has delivered nearly 5% more.

The annual capacity factors at four out of the six wind farms showed an upward trend with time.


The implications of lower-than-expected electricity production on the financial health of wind farms are important.

If a project yield is overestimated by 10%, the revenue from electricity sales will fall short by the same amount. However, rates of return are also important.

A paper by J Thomsen of Vestas at a financial conference in 2010 suggested that a 10% deficiency in energy production would reduce the internal rate of return of a typical project from 12.9% to 11.3%.

The implications for the equity returns would generally be more severe, but sample calculations suggest a 10% deficit could reduce the equity yields from 13% to 6%.

If wind speeds turn out to be 2% lower than anticipated, that could cut the equity returns from 13% to 10%.

However, it is important to note that the precise financial structure will influence the figures and that the numbers quoted are indicative.

Have you registered with us yet?

Register now to enjoy more articles
and free email bulletins.

Sign up now
Already registered?
Sign in