Models used by the wind energy industry may need to be adjusted, but ultimately, even more efficient windfarms will emerge after recent revelations by the world’s leading developer of offshore wind energy
Ørsted’s admission that it has been over-estimating production from its offshore windfarms because of the combined effect of the wake effect and a phenomenon known as blocking may have come as a surprise to many, but DNV GL raised questions about blocking 18 months ago after it conducted a research project. In fact, blockage has been known about for some time, but has only recently been researched in detail. A 2-3% impact on annual energy production was first discussed at EWEA 2014.
The results of DNV GL’s research project indicated that standard models used by the wind industry ignore the blockage effect, resulting in an overprediction bias that pervades an entire windfarm, not just individual turbines.
Until now, says DNV GL, energy production assessments (EPAs) only assumed a turbine can affect turbines located downstream of it. Any influence on turbines upstream or laterally has almost always been ignored.
DNV GL says this ‘wakes-only’ approach to turbine interaction and wake effects’ impact on turbines downstream are the only form of interaction in this widely used approach. Accordingly, energy production changes due to turbine interaction are usually called ‘wake losses,’ and models used to predict array efficiency are commonly referred to as ‘wake models.’
Now, however, DNV GL believes that a shift away from this approach is required, one that will require more evidence and a consensus understanding of how these interactions affect windfarm energy production.
“Compared with wakes, blockage effects, and the gradual wind speed changes associated with them, resist direct observation. Nevertheless, a growing amount of evidence from wind tunnel measurements, patterns of turbine production and measurements in the field have highlighted the pronounced influence of blockage,” it says.
The blockage effect arises from wind slowing down as it approaches wind turbines. There is an individual blockage effect for every turbine and, DNV GL research suggests, a global effect for the whole windfarm, which is larger than the sum of the individual effects.
In a recent webinar on the subject, DNV GL business manager Ioannis Papadopolous and head of renewable analytics Carla Ribeiro provided an insight into blockage, what it is and how it affects production. They stated that, although its effect is not as significant as wake effects – which can affect production by 20% – blockage could reduce production by up to 5%.
The problem is, they said, that it is simply not accounted for in traditional models.
DNV GL’s work indicates that wind speeds at upstream perimeter towers are lower than freestream and lower wind speeds mean the individual turbines just downstream produce less energy than current EPA practices would estimate.
Its work also suggests that a group of turbines on the upstream perimeter of a windfarm produces less than EPAs practices would estimate and that prediction bias on the upstream perimeter affects an entire windfarm. This means the magnitude of the overprediction bias in the wakes-only approach to turbine interaction can be significant and that power curves are probably an additional source of bias.
Ms Ribeiro says the magnitude of the blockage effect was affected by a number of factors. These include but are not limited to turbine density, rotor diameter (for a given hub height) and wind speed. Its effects are also greater in stable rather than neutral conditions.
The magnitude of the loss due to blockage is also likely to be highly site-specific, she says, “but it will affect projects large and small, onshore and offshore.”
Although the four windfarms on which DNV GL conducted measurements were all onshore, there is no evidence, she says, that losses due to blockage will be greater or smaller in offshore windfarms.
DNV GL’s experts conducted a series of measurements before and after the commissioning of a quartet of windfarms that clearly demonstrate that wind speeds just upstream of each windfarm decrease relative to locations farther away, after the turbines are turned on.
Their research found that at two rotor diameters upstream, the average derived relative slowdown is 3.4%; at seven to 10 rotor diameters upstream, the average slowdown is 1.9%. It used theory, measurements and computational fluid dynamic models to determine the cause of the phenomenon.
“Simulations point to windfarm-scale blockage as the primary cause of these slowdowns,” says DNV GL in a paper it recently published on the issue. “Blockage effects also cause front row turbines to produce less energy than they each would operating in isolation. Wind energy prediction procedures in use today ignore this effect, resulting in an overprediction bias that pervades the entire windfarm.
“Fluid dynamic interaction between turbines changes the power production at each turbine relative to what it would produce in isolation. The effect can be large and therefore must be considered when estimating energy production of a prospective windfarm.”
Despite the fact that a better understanding of blockage and its effects has resulted in the world’s leading developer of offshore windfarms having to admit it has probably been over-estimating production, Ms Ribeiro says DNV GL’s findings will be beneficial to the wind industry in the long-term.
“It is better to know about this,” she says, “so that industry can takes steps to mitigate the effect of blockage and optimise the layout of windfarms to minimise it.”
The good news is that DNV GL has also developed a blockage effect estimation tool (BEET), which is available to industry, although the tool is not site-specific. Derived from CFD work on a range of generic windfarm layouts, it is fast but is not a perfect tool as it could overlook some site-specific aspects of how blockage may affect production.
For its part, and as highlighted elsewhere by OWJ, Ørsted believes that, ultimately, understanding blockage and its effects will have benefits, and that solutions developed will further enhance wind power’s efficiency, cost-effectiveness and business case.
Grand challenges remain, scientists say
After OWJ first reported on Ørsted’s admission that it believes it has been over-estimating production from its offshore windfarms, DTU Wind Energy head of section, loads and control Katherine Dykes contacted OWJ to say she believes there are still “challenges around flow uncertainty” and “large machines” – such as next-generation offshore wind turbines – “are very much at the heart of it.”
Along with colleagues from DTU Wind Energy and a large, international group of wind researchers Ms Dykes recently published an article in Science that is very relevant in light of the current situation with Ørsted.
In the article, Ms Dykes and her colleagues around the world say additional research and exploration of design options are needed to drive innovation to meet future demand and functionality. “The growing scale and deployment expansion will, however, push the technology into areas of both scientific and engineering uncertainty,” they wrote.
They went on to explore various ‘grand challenges’ in wind energy research they believe must be addressed to enable the industry to supply much more of the world’s electricity needs.
Drawing on the findings of a recent international workshop, they identified three grand challenges in wind energy research that require further progress from the scientific community. These are: improved understanding of the physics of atmospheric flow around windfarms; materials and system dynamics of individual wind turbines; and optimisation and control of windfarms comprising hundreds of individual generators. “These grand challenges are interrelated, so progress in each domain must build on concurrent advances in the other two,” they say.
“Characterising the wind power plant operating zone in the atmosphere will be essential to designing the next generation of even larger wind turbines and achieving dynamic control of machines. Enhanced forecasting of the nature of the atmospheric inflow will subsequently enable control of the plant in the manner necessary for grid support.”