| The Energy Challenge - Wind Revisited | ||||
|
||||
Bird-kill
A review of 20 wind turbine bird kill studies over the last 15 years, with 9 from California, of which 5 from Altamont gives the following actual results:
From http://www.bpa.gov/Power/pgc/wind/Avian_and_Bat_Study_12-2002.pdf, table 4, after eliminating all studies that had less than 20 turbines (4 studies) and substituting Thelander et al 20031 for Thelander 2000 at Altamont, we have:
For the 11 studies outside California: average length of study = 1.68 years, number of turbines studied = 488, # of deaths all species = 250, # of dead raptors = 8, and from these numbers we can calculate that deaths/yr all species = 172 and deaths/turbine/yr = .35. For the 9 studies in California we find: average study length = 1.72 years, # of turbines studied = 3900, # of deaths all species = 1058, # of dead raptors = 418, # of deaths/yr all species = 615, raptors = 243, # of deaths/turbine/yr all species = .158, raptors = .062. The global unweighted average is only 0.18 deaths/turbine/yr.
However, WEST Inc., the company that produced this report, has weighted the results by MW produced (to better reflect current turbines), and has adjusted the counts for searcher efficiency and scavenger loss (both of which they have measured through controlled experiments) and have come up with an adjusted number of 2.19 deaths/turbine/yr, about 12x higher. A control performed to estimate bird deaths in similar areas without turbines strongly suggests that at least 1/3rd of the carcasses found around turbines probably died from other causes.. WEST also advise that adjustments for searcher efficiency and scavenging vary from 2x to 8x, a very wide range, and note that they have been very conservative (high) in their estimates to forestall disagreements. It seems likely that the 12x factor is near 1/3rd birds not killed by turbines and could still be on the high side. A factor of 8x would give an adjusted number of 1.4 deaths/turbine/yr. and would seem more likely. For the 12,000 turbines operating in the USA at the end of 2003, this would mean 17,000 mortalities per year. At least 20% of these are English sparrows, starlings and rock doves.
Seventeen thousand birds/yr. killed by wind turbines may seem like a large number, taken out of context, but consider that the range of estimates for all birds killed annually in the USA by cars, tall buildings, communication towers, power lines, pesticides, aircraft and domestic cats is from 300 million, to well over a billion. Even when we get to a hundred times more turbines than now, their bird kill will be in the order of ½ of 1% of the low end estimate for all other causes.
While bird deaths are regrettable, and steps are being taken to minimize the wind turbine contribution, wind turbines make such a trivially small contribution to the total death rate, that bird kill is simply not a reason to impede growth of wind generated renewable energy. There is no published evidence that any species is in danger due to wind turbines, and there is some evidence (at least at Altamont) to suggest that wind farms create environmental conditions that support growth of bird populations.
Repowering Altamont
Even as great an expert on wind power as Paul Gipe (2) has commented that repowering this wind farm will have no effect on bird kill because there is a strong relationship between swept area and bird kill and for the same power permitting there will be the same swept area. This assertion does not stand careful examination. What we need to consider is annual energy produced, not rated power.
From the report Paul reported on (1) Altamont was permitted for 580 MW, and had 7340 turbines in 1987/88, Assuming the same permittting was valid in the mid 1980s, (it might have been lower) the average turbine rating was 580,000/7340 = 80 kW. We know from a USA wind capacity study published in 1992 (3) that average efficiencies in 1991 were taken to be 25% and energy losses in the machinery were 25%. (That means that 25% of the energy in the wind passing through the swept area was captured by the rotor, and 25% of captured wind energy was lost in getting electricity into the grid.) Corresponding numbers in 1995 are about 35% efficiency and less than 15% losses and must be even better today. Those old turbines were also on low towers, didn't capture upper level winds and had capacity factors maybe 2/3rds of today's. Now consider replacing 25 of those 80 kW turbines with 1 of today's 2MW turbines (25x80 kW = 2 MW). The Eout of today's one machine will be at least 28/20(capacity factor) x35/25(efficiency) x85/75(losses) = 2.2times the Eout of the 25 old machines. This is a conservative estimate; a good case can be made for factor 3. In other words, if we ignore power permitting and look at energy we can replace 55 old turbines with 1 new turbine. The old turbines had typical rotor diameters of 19 meters on 24 m towers, for a swept area of 15,600 sq. m., for 55 turbines. One new 2 MW turbine will have a rotor diameter of about 84 m on an 80 m tower, for a total swept area of about 5540 sq. m., or 35% of the 55 turbines it replaces.
However, considering only swept area is probably too simplistic. In the above mentioned NREL study (1), considering only horizontal axis turbines that were studied for 25 months or more we find 560 units of which 185 experienced 1 or more collisions, and that:
While increasing swept area seems to generate a disproportionate increase in collisions, there is a hint here that at 33 m diameter the effect is already weakening, although the sample may be too small for real statistical significance. What could be happening here? Consider that a 17 m dia. rotor on a 20 m tower (typical for the Altamont) sweeps a band between 38 and 94 ft. above ground level, and a 33 m rotor on a 30 m tower 44 – 153 ft. Interestingly, California, with its serried ranks of old low fast rotating turbines seems to have a much lower death rate than the rest of the country for the all species number. A guess is that California has few or no turbines in migration flyways, or too low to interfere with migrations. Local birds on the other hand will tend to fly below 100 feet (personal observation), right where the swept areas of the old turbines are concentrated. Raptor concentrations may be higher in Calif., and raptors may be paying least attention when going after game below 100 ft., again where the swept area is located. Modern turbines in the 2 MW and up class have 80 to 110 m. diameter rotors on 80 to 100 m. towers. Their swept area has been elevated right out of the highly populated band at Altamont. It seems likely that fewer and much higher modern large turbines, carefully sited in California, would cut bird kill dramatically, especially for raptors. (Unfortunately it is probably precisely these high swept areas, reaching up into flyways, that is the problem in the rest of the country).
Another issue for Altamont has been the killing of golden eagles. The above mentioned study (3) notes that the only turbines that killed golden eagles during the 34 months of the study were located in canyons. If we ensure that none of the new turbines are in canyons the deaths of golden eagles should go close to zero.
Repowering Altamont pass seems desirable both for increased energy generation and reduced bird kill.
EROEI
It is hard to find any calculations of wind energy payback. Best results came from googling “energy payback” or “energy balance” comma wind. Most of the results were pre 1995, for turbines of 600 kW and smaller, lacked details of the analysis, and the older results are clearly doubtful (paybacks near 40x), so have been discarded. A 1997 Danish paper (4) concludes with a return of 80x for a 600kW turbine which seems clearly anomalous and so required some detailed analysis. The first hitch is that the 80x is for class 6 winds in a smooth (offshore type) environment, but the Ein was not for an offshore installation. They also gave a class 4 roughness 2 result as 60x, which fails the test of computation. The 80x is for 7 mps windspeed with capacity factor 27 and the 60x for 5.5mps windspeed with capacity factor 21. However, given the cube law, if 80x is OK for 7 mps then 5.5 mps will only give 40x, and 21/27ths will be 31x. (In such lower speed winds the turbine would no longer meet a 600 kW rating). The Danish analysis calculates energy to mine and transport coal, but does not include embedded energy, nor ash disposal. They also start with a power plant thermal efficiency of 47%. (The Danes use a lot of combined heat and power). They discount the 47% to 43.5% due to coal mining and transportation energy requirements of 8%. If we start with the USA situation of about 33% power plant efficiency, and discount that by 12% to include embedded energy and waste disposal, we have 29% efficiency vs the Dane’s 43.5%. The resulting increase in Ein decreases the return to 21x, which seems reasonably likely.
In order to get a good idea of wind EROEI, one of the variables that has to be dealt with is “capacity factor” or % equivalent full load or full load factor. The wind industry uses the term capacity factor. The problem with analysis is that the realized capacity factor is generally lower than the designed capacity factor. It is very hard to find more than anecdotal information, but it seems that most on shore USA installations were designed for about 33 to 38% capacity factor. Total USA averaged 24.6% in 2001, 26.7% in 2002 and 28.7% in 2003. The most likely reason for the increase is better load matching as part of the learning curve, with some contribution from newer, larger turbines. However it seems that one should discount designed capacity factors by about 20%.
A 1998 U. of Wisconsin study (5) gives energy return results for 3 windfarms as 17x, 23x and 39x. The 17x is for a 2 turbine location with an exceptionally high tower, heavy nacelle and low economy of scale and should be discarded. The 23x is from measured Eout. The 39x is theoretical for 143 750 kW turbines, based on a designed capacity factor of 33 and a 25 year expected life. If the capacity factor is discounted 20% and a 20 year life is assumed (as in the Danish paper) the payback becomes 25x. This study didn’t account for embedded energy or energy to mine and transport coal, so this number should be adjusted down another 10% to about 22x which seems consistent with the corrected Danish result.
A 2002 Vestas paper (6) provides a life cycle analysis for 2 windfarms (one offshore and one onshore), using 2MW turbines. They include a theoretical energy balance calculation that gives a return of 27x for the offshore farm and 31x for the onshore farm. (The offshore farm includes transformer and transmission line to shore.) Adjusting Ein by 5% for embedded energies and some acknowledged subcontractor lacunae, and Eout by 20% for capacity factor we get 20% and 24% respectively, which is again consistent. They do point out that the offshore farm is likely to have a useful life at least 50% longer than the 20 year spec’d, due to smooth winds, so the return might be more like 30x.
What about Odum?
Evidently Odum came up with an EROEI of 2x for wind based on mid ‘80s data. To update his work we must first think about the last 20 years’ progress in various fields like electric arc steel making, low waste casting, computer aided design, numerical control machining, multi-axis machine tools, laser cutting, fibre-glass prepreg, just-in-time(JIT), TQM, lean manufacturing, etc. If we were still making little wind turbines like 20 years ago, how much less Ein per unit would we use today? Since energy per unit of GDP has better than halved since 1973, we can be quite confident it would be less than 75%, already improving Odum’s EROEI by >1.33x.
As noted above, we can now replace 55 early 1980s turbines with one modern 2 MW turbine, with the same Eout. That's also replacing 55 mounting pads, 55 towers (albeit with one larger tower), 55 grid connections. How much less energy would it take to cast and machine one large gear set than 55 smaller gear sets, wind one large generator than wind 55 smaller generators, manufacture 1 set of bearings than manufacture 55 smaller sets of bearings, mold one set of rotor vanes than mold 55 sets of smaller rotor vanes, etc? How much less energy does it take to service one modern low service turbine than to service 55 old high service turbines? Overall a factor of 7 to 10 would seem believable Using a factor of 8 and combining manufacturing efficiencies and turbine count we have a reduction of Ein by >factor10 for the same Eout. If Odum’s EROEI for wind of 2 was right circa 1985, we have to accept an EROEI of >20 today, using "the same methodology", again consistent with other estimates.
Future EROEI progress?
None of the referenced studies consider that only the nacelle and rotor require replacement after 20 years. Using the Danish study details for the portion of Ein charged to nacelle, rotor and maintenance, and assuming one replacement with a 40 year life for all other Ein, raises the return by 33%. Again this is a conservative estimate as modern rotors are designed for a minimum 20 year life.
Finally we need to consider efficiency. The Betz limit on efficiency is 59% and the assumed upper practical limit has been taken historically as about 43%, mainly due to aerodynamic considerations like turbulence, especially at the rotor tip. 1995 data gives efficiencies near 37% close to designed windspeed, and near 35% over a range of windspeeds not too far from design. Mention has been seen of efficiencies above 40%. A 2004 result from Enercon of Germany (7) claims 56% efficiency for one machine with a new blade design and winglets on the rotor. That is probably an optimum number, but it holds promise of 45% or better in production, which could raise returns by another 20 to 30%.
It is probably safe to say that EROEI of modern wind turbines in class 4 wind areas in America, with an assumed 20 year life is at least 20 to 25x. With one nacelle/rotor replacement this would go to 26 to 33x. With optimized blade design and winglets, the future expectation can be > 40x. If expected working life can be extended to 30 years then an EROEI >50x is a realistic goal. This is much better than we can ever expect from fossil fuels.
Conclusions
Bird kill due to wind turbines will never be more than noise level compared
to other causes of bird kill.
There is no evidence that wind turbine bird kill threatens any species.
Altamont should be intelligently repowered, using modern technology, to
provide more energy with lower bird kill.
Odum may well have been right in his two decades old estimate of wind EROEI
as 2x. EROEI in 2005 however is surely better than 20x, and probably better than
25x, with promise of exceeding 50x in less than a decade.
Utility managers and financiers need to know the facts when facing disinformation from renewable wind energy opponents.
References:
(1) http://www.nrel.gov/docs/fy04osti/33829.pdf
(2) http://www.wind-works.org/articles/NRELBirdReport04.html
(3) http://www.nrel.gov/wind/wind_potential.html
(4) http://www.windpower.org/media(444,1033)/The_energy_balance_of_modern_wind_turbines%2C_1997.pdf
(5) http://www.ecw.org/prod/180-1.pdf
(6) http://www.vestas.com/pdf/miljoe/pdf/LCA_report_efp_170105.pdf
(7) http://www.earthscan.co.uk/news/article.asp?UAN=24&SP=332572698817342720322&v=3
To join in on the conversation or to subscribe or visit this site go to: http://www.energypulse.net