There’s a debate happening in the wind industry that has serious implications for the success of projects: How high should wind measurement towers stand in order to be accurate?
For example: If turbine blades extend up to 120 meters high, why do developers use 60-meter towers to forecast power output and project revenues? And how accurate are those projections from short towers up to hub height?
Such questions were the focus of a competition designed to test the accuracy of wind projections. The competition, in which six analysts were charged with comparing results to see who was the most accurate compared to actual 80m data, was part of AWEA’s Wind Resource Assessment Workshop held last September.
The results of the forecast were incredibly varied – but they offer some good insight into how data is collected and analyzed, as well as the accuracy of measurement tools.
Before we get to the contest, let me list some benefits and drawbacks of short towers:
A 60-meter tilt up tower is cheap to instrument and erect, making it more attractive. A tower constructed at hub height or higher costs more than twice that of a short tower. The geometry of a 60-meter tower is cheap and simple: it’s a tall pipe with lots of guy lines. An 80-meter tower is a climbable, guyed tower with a large, poured concrete base – and that simply costs more.
A short 60-meter tower probably costs $45,000 over the life cycle for all costs. An 80-meter lattice tower costs more than $85,000 over a two-year life cycle, which includes:
- The tower itself
- Pouring concrete and other construction costs
- Erection of the tower
- FAA permit application and processing (required for towers over 60 meters)
- Site acquisition
- Local building and other permits
- Removing the tower at the end (another $15-17,000).
Of course permit applications for any new tower awakens the “Not in My Backyard” crowd as well.
While shorter towers have up-front cost advantages, their lower accuracy is less attractive in this capital-constrained period. Project finance investors have historically been willing to open their wallet and back projects with thin due diligence. But that’s no longer the case as credit standards have tightened. A shortfall in the actual output of recent wind projects has given cause for more scrutiny than in previous years.
Technically, shorter towers are inferior as well. A 60-meter mast covers just 35% of the blade sweep of a typical turbine – say, for example, GE’s 1.5 MW unit on an 80-meter tower. Estimating power over the entire blade sweep and the 65% that is not covered with actual data is a guess. And wind speed data from 60-meter towers needs to be projected up to an 80-meter hub height or higher.
The professional challenge for meteorologists is to merge the physics of fluid dynamics with the chaos of weather. There’s a technical term for these projections: Guesstimates!
The guesstimate takes actual 60-meter tower data on wind speed (average, minimum, maximum, standard deviation) and uses an industry standard Power Law formula to project it up higher.
Five out of the six analysts competing at the AWEA Workshop used a Power Law (one used the Log Law) to extrapolate to hub height. Wind speed almost always rises with altitude, so the actual wind speed at 80 meters is usually higher than the 60-meter data. The right coefficient can project data up. But suppose each site needs a different coefficient? There goes the standard!
The coefficients used to project short tower data up to the hub height and beyond are not immutable laws of nature.
As V-Bar’s Greg Poulos said: “Atmospheric winds are non-linear and chaotic.”
Applying the formula to a specific data set is tricky.
To quote John Vanden Bosche of Chinook Wind at the conference: “A logarithmic profile overestimates wind below hub height and underestimates wind above hub height. With non?logarithmic profiles, strange results can occur.”
And the same contest found analysts projecting capacity factors almost 2% higher than actual measured data would indicate.
“Most of the analysts over-predicted output,” said Kevin Walter of Tradewind Energy.
V-Bar’s Poulos expressed caution about the limits of modeling vs. validation with actual tall tower data, saying: “Modeling as applied to project wind energy resource assessment is not a substitute for tall tower meteorological data.”
Hub height projections are driven by some conventional metrics on how wind speed rises with height and how the surrounding terrain affects wind speed. Yes, wind speed almost always rises with height – but not always.
The variation in data from 60-meter towers and the difficulties in constructing larger, 80-meter towers presented an interesting opportunity for business. Two years ago, WindPole acquired met data acquisition rights to 12,000 existing tall (80+meter) towers. These comprise about 85% of the available communication towers in the US.
WindPole has actual data from hub-height met towers in ten states which show that projecting from short towers yields a forecast range broad enough to alarm risk adverse investors. A broader range of possible project yield means that investors face uncertainty. Uncertain investors will require more entrepreneur equity in a project, and will offer less debt. This means that short tower data ties up more developer equity and possibly reduces the size of wind projects.
(Here, an 80-meter tower raises the P-99 NCF by a full point, thus gaining the developer an additional $3.58 million in project debt).
Does short tower data average out as accurate on a national basis? Maybe, but that’s not an acceptable answer to an investor in a specific project. The investor wants to know the forecast over the 30-year life of a project. They look for the range of possible down years and the P90 forecast that has a 90% chance of being achieved. On most projects, a P90 value is substantially less than the optimistic output forecast embraced by the confident entrepreneur.
Developers are optimists; focused on the P50 (50:50 chance the winds are higher/lower. Investors, however, are pessimists. They want to know wind speed and power output with 90% assurance or higher. And guess whose risk profile and metrics are reflected in the? Yep, the investor’s lower P90.
To solve the short tower problem, the wind power sector is slowly adopting tower technology that is standard in the communications sector. This means Sabre or Rohn towers typically used for broadcast antenna or cellular mobile radio.
These tall towers will eventually be required for met data on all wind projects along with a mix of short 60-meter towers, and maturing technologies like SODAR and LIDAR. For a large wind farm, all of these wind resource assessment costs, while painful, and upfront, pale in comparison to the cost of turbines, towers and the balance of plant.
We expect financing requirements for wind resource analysis to get tighter as the industry matures. As investors trend toward requiring more supporting data, more due diligence, and more de-risking of projects, tall tower data (and even better, full blade sweep data) is a simple solution.