"Wind power is still one of the most scalable renewables, but our research suggests that we will need to pay attention to its limits and climatic impacts if we try to scale it beyond a few terawatts," said David Keith of Harvard University.

Along with Amanda Adams of the University of North Carolina, Keith used a parameterization of the atmospheric effects of wind-turbine arrays developed for use in a mesoscale model, in this case the Weather Research and Forecasting Advanced Research model (WRF-ARW). By running simulations for turbine arrays 2.7, 30 or 270 × 103 km2 in size, the pair found that wind-power production would saturate at around 0.5 to 1 W m2 as turbine density increases.

In a development that Keith finds "surprising and interesting" in a field that has seen "quite vigorous disagreements", these numbers correlate well with the most recent estimates of wind power that use global climate models. "It's one of the things I'm really happy about," Keith told environmentalresearchweb.

Earlier estimates of global wind-power capacity, which ranged from 56 to 400 TW, simply added up the best estimates of local wind. "At this point even though there are ways in which we and the [Mark] Jacobson group disagree, we both agree that you can't do that – you have to take account of these distant effects," said Keith. "Infinitesimal things happen a long way away. You can see footprints halfway around the globe from a 10 TW or 5 TW wind-turbine array."

By ignoring the effects of wind-turbine drag on local winds, these earlier estimates assumed that 2–7 W m2 of wind power can be produced over large areas. But Adams and Keith's model indicates that, for wind farms larger than about 100 km2, wind-power production is limited to roughly 0.5 W m2.

"If that's right then...it's still true that the upper limit of wind power if you're willing to cover the entire world with wind turbines is very large, say 100 TW, so bigger than matters in policy terms," said Keith. "But if you had to do something more realistic, where you select where you're going to put wind turbines based on being relatively close to human energy demand, and you had to find relatively good wind resources and not have a big environmental impact, then you're down to a much smaller number." And that might be the point at which the constraint of half a watt per square metre becomes important.

"I don't think it matters tomorrow but remember we need maybe 30 TW of carbon-free energy by mid-century if we want to cut carbon emissions towards zero," said Keith. "So if wind power is going to supply a significant fraction of global carbon-free, you're talking 10 TW. There are plenty of advocates of wind power who are talking numbers that big."

As total wind-power generation approaches a few terawatts, there will be "both a climate impact and meaningful constraints on the amount you can extract". Productivity at a given site will be less than if the other sites had not been there, Keith explained. "You'll begin to edge towards saturation."

Once you stop assuming that each wind-power site is independent of the others, calculating global wind capacity becomes very hard. "It's an optimization problem, assessing where to put them [wind-power sites] optimally, and nobody's done that research yet," said Keith.

As plans to build hundreds of gigawatts of wind power develop, Keith believes there needs to be some form of collaborative planning process that looks at the impact of one turbine farm on other sites. "We've got some traction with this parameterization we've built into the WRF weather research and forecasting model," he said. "Some of the forecast meteorology centres – NCAR or the meteorology centre – should begin to do these simulations in a more operational way."

The team's parameterization also indicated that wind farms could increase temperatures in the lower part of the boundary layer by reducing wind speed, which alters the natural wind shear and boosts turbulence, mixing higher-temperature-air downwards.

Now Keith plans to run a Harvard–MIT meeting to bring together "a bunch of people who've had different opinions about this to talk about what needs to be done next".

He reckons the next challenge is to bridge the scale boundary. "We need to build slightly better parameterizations in the high resolution model," he said. "You could even go a scale down; it would be nice to go from the scale of real wind-turbine arrays and direct measurements, to tie that scale better to the one we're looking at in this paper, and then you could tie that scale to the global model scale. It's a year of student or postdoc time, but it would be really worth tieing these findings more firmly to observations."

Related links

Related stories