This is part on an ongoing series of diaries (1) where I explore energy policy, currently Wind Power, the fastest growing electric power source in the world. After a break to do Maple Sugaring (an average year) I'm looking at this again. How much does it cost to keep Wind Turbines going? How long will they last? These are questions that keep getting raised over and over again.
Capital Costs have been based on a 20 year lifetime. Is that realistic? Let's look at the lifetime of turbines based on 25 years of data from Denmark. This shows we haven't hit any real lifespan for decent size (1 MW or greater) wind turbines. The jury is still out, but with 650 Wind Turbines of 1 MW or greater with an average age of 7 years there have been 10 decommissionings, 4 of those were replacements with larger turbines and several more look like tests. I could only find 2 unknowns. This isn't enough to usefully extrapolate a lifetime model, but it does put a cap on things. Even dropping to 500+ kW size gives only 17 decommissions (including the 10) out of over 3500 turbines (0.005 of them) with an overall average age of 13 years. Unless there is a cliff at 17 years or something it looks like 20 years is a low end estimate.
A while back a contributor here at Kos posted a diary with a link to the Danish Wind Data (2) which is quite interesting. He tried to show that lifetimes for Wind Turbines are about 15 years by looking at the data in a quite, well, unusual and incorrect way (you cannot discount all the still working turbines when you calculate a lifetime... and you should look at sizes) but he gave the source and how he did the calculations and that was a valuable thing. I thought I would look at this same source of information and try to come up with a failure distribution model and really get a lifetime. I used to do that for a living, but when I actually looked at the data from Denmark, it became clear that it wasn't going to be too useful. There are thousands of turbines deployed and thousands of turbines decommissioned from the 1980's forward in the spreadsheet, but looking closer something jumps out. There has been a large technological change in Wind Turbine deployment in the last 5 years or so. The size and efficiency of turbines has jumped considerably. I am interested mostly in the turbines that are of the size used for utility level wind projects in the U.S. right now and that means generally 1.5 Megawatt capacity and up. I thought I would just look at 1 MW generators and up in Denmark and use those statistics.
Well first of all, the Median size of the generator that had been decommissioned was 90 Kilowatts. A size not seen much in the U.S. nowadays in commercial wind projects. Even the third quartile size was only 150 kW. Even though these might be interesting they are not the same technology as the larger turbines and are quite different in design. Looking at the larger ones, there are 650 turbines with over 1 MW capacity that were in use or had been in use in Denmark with an average lifespan (to today) of seven years and there were exactly 10 decommissions.
Well that is possibly an interesting number, greater than 1 percent, which would indicate a long lifespan, but is still not enough to get decent estimates. But Not so fast. I looked at the individual items. 6 of them were all right next to each other (less than a couple of kilometers from end to end of the six) in a line of three different manufacturers. Lets look individually. I'll use the last 6 digits of the Unit ID as the tag:
62643 NEG Micon 1.5 MW active from 5-Dec-02 to 4-Sep-03 replaced on same location by
62766 NEG Micon 4.2 MW ! (110m rotor? wow) active from 1-Nov-03 to 21-Nov-07
62650 Vestas 3 MW (90m rotor) active from 12-Nov-02 to 15-Nov-04 replaced on same location by
71096 Vestas 3 MW (100m rotor) active from 4-Dec-04 to 9-Nov-08
62667 Bonus 2.3 MW active from 22-Dec-03 to 15-Aug-04
62681 Vestas 2 MW active from 19-Dec-02 to 24-Nov-04 this is a about 100 meters from the other Vestas above
This looks like some kind of evaluation area where they are replacing units and trying them out. Three different types next to each other of varying types in a farm is very unusual. The sizes are interesting also. I don't think we can use these is failure statistics until more information is given.
35678 Vestas 1.5 MW active 8-Jun-88 to 28-Jan-02 replaced by 2.5 MW unit same location.
35692 Bonus 1.0 MW active 1-Oct-94 to 31-May-08 replaced by 2.5 MW unit same location.
So these two were upgrades to bigger units. I don't know if there was a failure or this was an upgrade. That wouldn't help with lifetime data since economic choice doesn't count for a cost – it is a different cost calculation than for lifetime cost. One that is interesting but different.
103070 Bonus 2.3 MW active 30-Sep-02 to 18-Jul-05 “offshore” unit.
Well that is a type that is out there so it might be valid although the others of this type are land versions. Don't know what the offshore means but it is the only one with this designation, so again it is unique.
35685 NEG Micon 1000 MW active 8-June-88 to 1-Jul-99
So we have one 11 year lifetime unit.
You may take this as good news for long term wind viability and it is in a sense that the lifetimes are not catastrophic. But this little data doesn't give enough for decent conclusions, except that these units are probably still in a “bathtub” trough for failures. That is, these few examples, even taking them all as failures do not constrain a failure model enough to give decent estimates. In a Weibull (bathtub looking) failure distribution - which is often applicable to equipment lifetimes - there are a few teething failures, then few failures for some time, then as the natural lifetime approaches, an increase of general failures, Using this data I could pick parameters that would fit lifetimes up to 100 years. That is meaningless, something would wear out before then. So we are still in the "trough" for current tech wind turbine lifetimes.
To check, if we look at turbines of 500 kW capacity or more we get only 17 decommissions (including the 10 above) out of 3245 turbines with an average age about 13 and a half years. Well, this supports the idea lthat these things last a long time, but these 500 to 900 kW turbines are not wholly representative either.
In any event this data is certainly consistant with the 20 year lifespan for financial calculations. It is consistant with longer lifespans also. We'll have to keep watch as the current technology ages in use. A longer lifespan is directly correlated to cost per watt for wind since capital cost is such a factor.
Component wise, the “engine” itself needs refurbishment every 7 to 14 years. That estimated cost is built in the $0.02 operating cost per kilowatt that is currently used. This does depend on the cost of turbines which over the last year have risen drastically. Even so, this is still in line with planning projections. Things like the lifetime of the foundations are also given as 20 years, but that is not true. There is no reason that they won't last 50 years or more, with reasonable care. Wind turbines don't put that much different load on their base than a tall building, and those foundations can last a century or more. The turbines are much lighter but would have more wind load, obviously. There is active work on the costs of these, and it is a well understood area of engineering, so I expect some improvement over time.
It will be important to watch the ongoing costs as the turbines near their nominal lifetime. When will the operating costs become high enough to warrant rebuilding? Many people are still behind the curve on Wind power not realizing how quickly we are deploying it at scale.
Next time I will look at Wind's ecological costs. Mainly habitat invasion, bird and bat kills. These need to be watched but relatively are less than most other human activities.
(1) Wind By the Numbers 2009 and Beyond, Money
(2) Danish Energy Agency Wind Turbine Master Data Register