– Dr. Alex Cannara (advanced degrees in math and engineering)
Since wind needs lots of elbow room, determining the actual land requirement is an important bit of penciling. We call it “land density” – the amount of land needed per X amount of energy produced.
Since a wind turbine has a tiny footprint (the base of the tower) but needs lots of elbow room for its blades to catch a fresh breeze, land density is the proper method of wind farm land estimation.
And while it’s true that a wind farm can be planted in a wheat field, the greater truth is that wind companies are harvesting wind, not wheat. So the rationale of wind’s “real” footprint being just the base of the towers, is misleading.
One of our gimmes for onshore wind is that we accept that 70% of it would be on sweet spots in the midwest. But the remaining 30% would take up far more land than an entire national nuclear footprint.
Although not directly stated in the Roadmap, its land density for wind can be derived as 0.089 km2 / MWp.1 In an email we received from Dr. Jacobson, he confirmed the 0.089 value we deduced from his paper.2
That number is about one-quarter of NREL’s estimation of 0.345 km2, from their 2009 survey of U.S. wind farms.
In other words, the Roadmap claims that wind can produce 4X the power per square kilometer that NREL says it can.
Part of this wild discrepancy is because NREL took into account the irregular terrain (creek beds, odd property shapes, etc.) they encountered in their analysis in 2009. Another part is due to the fact that wind technology has improved.
Even so, the only way the Roadmap’s number could work is by siting their wind farms on flat, ideal, wide-open spaces.
To be more than fair, we’ll go with the Roadmap’s density estimate for onshore wind. But as dense as the Roadmay can lay out its farms, it still needs nearly 6X the land for its PV solar, this is still nearly 6X the land required for PV solar, which weighs in at a svelte 0.016 km2 per MWp.3
While the Roadmap’s land for PV solar is equal to Maryland and Rhode Island (literally every square foot of land in both states), its land estimate for onshore wind is a bit larger than New York State, while its area for offshore wind is a bit larger than West Virginia.
If you trust NREL’s land-for-wind estimates better than the Roadmap’s (as we do), figure on New York, Pennsylvania, Vermont and New Hampshire for onshore wind.
The Roadmap calls for 1,701,000 MWp of onshore wind. With an expected capacity factor of 28.9%, onshore wind will generate 492 GWs average, or 30.9% of the 1,591-GW grid.4
By 2015, we already had 73,400 MWp installed. So we’ll need to fabricate (and / or import) 1,627,600 MWs more.
Their plan is to use the new humongous new 5-MW turbines. That comes to 325,500 more onshore spinners.
With more moving parts than a solar panel, an onshore wind assembly will only last 10–25 years before the propellers and the mechanical contents of the nacelle
(the housing atop the tower) will need to be refurbished. But the tower and foundation should last as long as a reactor.
In 2014, the U.S. DoE (Department of Energy) estimated that 2-MWp wind assemblies cost about $1.71 per installed Wp (peak watt).5
Factoring in economies of scale and engineering advances, we’ve applied a 20% discount to the fabrication and installation of the Roadmap’s proposed 5-MW monsters. So anticipate $1.37 / Wp for wind during the buildout.
Two turbine refurbishment companies, one in Europe and one in the U.S., have told us that a complete overhaul, including rebuilding or replacing the blades, averages 10% of initial cost.
For a 60-year lifespan of the Roadmap’s onshore wind:
Onshore wind has the best value of any WWS system in the Roadmap: 30.9% of total energy, for 17.1% of total cost. A nearly two-for-one bang for the buck.
A 2011 IRENA study (International Renewable ENergy Agency) concluded that offshore wind is twice the cost of onshore systems.6 That’s $2.74 per installed Wp, with the same 20% discount that we applied to onshore.
The Roadmap calls for 780,900 MWp of offshore wind, with a brisk capacity factor of 38.8%, that would generate a yearly average of 304 GWs, or 19.1% of the grid.7
But in the harsh marine environment, mechanical equipment and blades only last about 10–15 years. So three overhauls, not two, will be needed over 60 years.
We have generously ballparked the overhauls at just 10% of initial cost, the same as onshore wind, even though offshore turbines are serviced at sea.8
Though it’s not the same bargain as onshore wind, offshore still has more bang for the buck than solar: 19.1% of total energy, for 18% of total cost.
Wind will comprise 50% of the Roadmap’s 2050 grid: 30.9% from onshore, and 19.1% from offshore, for a 60-year price tag of $5.33 Trillion.
That’s half of the grid’s total power, for one third of the grid’s total price. Pretty good deal. In fact, it’s tempting to just power the entire grid with wind alone. But there’s a catch:
While onshore wind is the best bargain on the menu, the amount of land it gobbles up makes it utterly impractical as a silver bullet to power the entire country, even if we could afford the energy storage – which we can’t.
Even with a wind density factor of 0.089 km2 / MWp, it would take 489,600 km2 to generate the Roadmap’s entire 1,591-GWs average with onshore wind.
That’s acreage equal to Florida, Georgia, South Carolina, and half of North Carolina.
Sorry, that just ain’t gonna happen. Even way out west.
See table 2, row 1, columns 2, 6 and 8. U.S. land area is taken to be 9.162e6 km2.
1.5912% X 9.162e6 km2 = 145,800 km2; 5 MW X 328,000 turbines = 1.64e6 MW;
145,800 km2 ÷ 1.64e6 MW = 0.089 km2 / MWp assumed land density for wind.
Dear Dr. Jacobson,
I am writing an article on renewable energy and need some clarification.
For onshore wind the 100% clean and renewable WWS all-sector energy roadmaps for the 50 united States shows 1.59% of US land area needed for spacing of new plants /devices. I take this to mean the entire area of a wind farm, what the National Renewable Energy Laboratory defines as Total Wind Plant Area in their 2009 technical report – Land-Use Requirements of Modern Wind Power Plants in the United States.
NREL defines Total Wind Plant Area as “the total area of a wind power plant consisting of the area within a perimeter surrounding all the turbines in the project”. [p.4, Sec. 2.2]
172 large wind projects were evaluated in the NREL study, obtaining a clear specification of the Total Wind Plant Area for 161 of them [p.10, Table 1]. Their combined Total Area was 8778.9 km², with combined generating capacity of 25,438 MWac, giving an Average Area Requirement of 34.5 ha /MW, or 0.0345 km² /MW, shown at lower right in that table.
The 100% WWS Roadmap, Table 2, states a target value of 1,701,000 MW, with 3.59% already built as of 2013. New buildout would therefore be 1,640,000 MW.
With NREL’s land-usage for actually existing large wind farms at 0.0345 km² /MW, the new land area required would be 565,800 km². That land area represents 6.18% of all US land, if Alaska is counted. This is about 4X greater than the 1.59% value for onshore wind in Table 2 of the Roadmap.
Perhaps the word “spacing” in Table 2 does not really refer to the Total Area occupied by large wind farms built in the US. Perhaps it refers instead to a theoretical model for flat land only, assuming a rectangular field with a turbine array spaced about 3 to 5 blade-diameters apart “sideways,” and 10 diameters apart in the direction of prevailing wind.
Under that assumption, analysis models anticipate land usage of 0.13 to 0.20 km² /MW [p.15 of the NREL report]. The center value of that predicted range, 0.165 km² /MW, would yield new land requirement of 270,600 km², or 2.95% of total US area. Even this idealization is substantially greater than the 1.59% of US land area specified in Table 2.
Could you help me reconcile these discrepancies?
Thank you in advance.
On 10/21/16 1:17 AM, Mark Z. Jacobson replied:
Yes, I will address this below. Also, I checked out your “Critique” of our U.S. plan, and while I am flattered you have taken such an interest, I would suggest you go into the spreadsheets more to see exactly how things are calculated. For example, you claim that the U.S. average capacity factor of wind and solar applied to our generation capacity give a slight underestimate of our annual power output but you omit the fact that we are including offshore wind in our 2050 mix (none of which existed in the U.S. at the time of the report), and CFs are higher for offshore wind than onshore, and you averaged wind and solar CFs and different types of solar CFS, then multiplied an average number by a total capacity rather than multiplying individual CFs by individual capacities and summing the results. Also, you used recent values rather than 2050 values, which we use.
With regard to wind turbine spacing areas, we use the standard metric for wind turbine area requirement Area (km2) per turbine = aD x bD where D is turbine diameter (km), and a and b are constants representing the sidestream and upstream distance between turbines in an area. For onshore turbines, we used a=4, b=7 and for offshore, a=5 and b=10. For the 5-MW, D=126 m turbine we used, these translate into 0.44 km^2/turbine (0.089 km^2/MW) and 0.79 km^2/turbine (0.159 km^2/MW), respectively.
A recent study that will be published shortly by an independent group analyzing the spacing of more than 1000 operating turbines covering 44 onshore and offshore wind farms around the world found that the mean distance between turbine towers was 4.2D, giving an approximate mean area of turbines as A = 4.2D x 4.2D = 17.6 D^2, which is much less than what we used (28 D^2 and 50 D^2).
In other words, the spacing areas we estimated are larger than spacing based on real wind farm data (thus our results are conservative), which is opposite from the conclusion you draw from the NREL report.
There are three reasons for this.
1) NREL does not provide any calculation of actual average distances between turbines towers, which is the relevant method of performing this calculation because the reason turbines are spaced is to avoid interference of the wake of one turbine with the next. It is irrelevant to know the irregular outside perimeter of a property based on project applications (which is what NREL used), particularly since the outside may be far away from the last turbine actually installed or could lie in a creek bed far away from any turbines.
2) The NREL report acknowledges on page 15 that their method of calculation “Wind Plant Area” results in overestimates and gives several examples why.
3) On page 4 of the NREL report, they further acknowledge that the Wind Plant Area is “subjective in nature” and “the total area of a wind power plant could have a number of definitions.” In their case, they define it based on project applications, which results in several of the overestimates given in (2) above.
On the other hand, the method based on data I described above relies on analyzing actual distances between turbine towers.
In sum, I believe our estimates overestimate rather than underestimate spacing area requirements based on real data.
This result is common sense as well, particularly as we go toward 1.7 million turbines in the U.S. Wind farm operators have an incentive to squeeze turbines as close together as possible to minimize transmission costs and land impacts, sacrificing some loss in capacity factor due to more interference.
Search for “NREL’s 0.029 value becomes 0.016”.
See internal footnote No. 40. See also:
Wind Energy Feb. 2017, Volume 20, Issue 2, pages 361-378.