Supply Chains Blowin’ in the Wind
September 18, 2009
This will be a very UK-centric post, but there are things that you could take away to apply to other countries if you wanted to. Be aware though that the data that I present here is not peer-reviewed and it is only semi-quantitative. I’m not an industrial electrician and I have had no access to detailed designs of wind farms. I asked several wind farm developers but had no success. For these reasons I’m going to call this an order of magnitude, or pathfinder study to be used to identify areas for further research rather than providing definitive ‘answers’.
It has deficiencies. I know what those deficiencies are. Don’t moan at me.
I’ve rounded everything in this post to nice figures. The nasty ones stay with me unless someone makes a specific request, but please remember, this is order of magnitude stuff. I think that I’m in the right ball park, but those with the hard data don’t want to pitch in (to stretch the baseball analogy).
I have tried to come up with a figure on the total amount of copper needed to satisfy the current UK policies on wind power, and while I’m sure that my findings are not news in the wind farm development community they may be interesting to others.
The UK’s wind energy strategy is to install around 37GW of new wind powered generating capacity over the next 10 years to 2020, and about another 10GW in the following 10 years. But we’ll deal with the first 10 years only because the degree of technology learning will be quite substantial and there are several disruptive technologies at the pre-commercial stage right now, especially in the field of high temperature super-conductors.
If we assume that the average size of conventional wind turbines installed through the next 10 years is 5MW, we’re looking at 7,400 new turbines. Some will be bigger, some smaller, but 5MW seems like a good place to start since there are several prototypes that size already working in Europe.
25GW (5,000) will be installed offshore, 12GW (2,400) onshore. This is the rough estimate put forward in the UK’s Low Carbon Strategy, but obviously subject to commercial realities.
Yes, I know that those onshore are likely to be restricted in size due to the planning regime and that bigger turbines may be developed offshore, but this is an order of magnitude study remember. It doesn’t actually matter that much how many turbines there, minimum safe copper usage is broadly proportional to the power that it is being used to generate and conduct (though for commercial, safety and regulatory reasons the actual engineering may break that proportionality, but again no-one was willing to tell me by how much).
So first things first.
How much copper is there in a wind turbine ?
Quick answer, just under 2 tonnes per MW of nameplate generating capacity.
I’ve used 4 peer-reviewed papers to get this answer, which I realise is quite a low number of data points and I’d like to get more data on the copper consumption of large modern turbines, but there are only a few peer-reviewed papers on this and some of them don’t have the data in a form that is usable. The largest turbine that was included here was 2MW, which is an other issue.
That means that just in the tower and nacelle each and every one of our 5MW wind turbines will contain around 10 tonnes of copper.
Sounds a lot ? A typical 5MW nacelle (with rotor, generator, gearbox and transformer) will weigh 150+ tonnes in total, so its really not that much in context.
We have our first estimate – 74,000 tonnes of copper embodied in the turbines. Easy.
Of course technology is moving fast within the mechanical portion of turbine development, with gearbox-less models, generators with super-conducting coils, generators with dynamic numbers of coils, etc, etc. So we can reasonably expect that figure to drop somewhat, even over just 10 years.
However, a wind turbine siting on its own is no use to man nor beast. What it needs is a connection to the grid. Cabling is a less dynamic technology. In fact cables haven’t really changed much in the last 50 years, despite that massive electrification of the industrialised nations during that time. Hell, there are probably cables in my street that are almost that old ! Don’t get me wrong cable performance has increased, but not really in terms of the amount of copper used, more in the engineering surrounding the conductors in terms of resistance to damage or corrosion, or lightness. In this field aluminium conductors don’t really get a look in.
In my workings I used the ‘off-the-shelf’ wind farm cable of choice, the Nexans 33kV (submarine version) for connecting both onshore & offshore turbines. Since what I’m interested in is the mass of copper conductor not the engineering around the cable, the actual version is not very important and Nexans provided the most complete datasheet.
At this point I will say that my methodology breaks down in large onshore wind farms with complex cabling topologies. Since no-one will show me their designs or costings, I have had to assume a ‘least distance’ method. In other words I’ve optimised for cable length, not wind farm cost. What my results show is that this is a reasonable assumption for onshore wind farms of about 30 turbines or less. Above that the cost of system elements, such as the sub-station, and the installation costs appear to start to exert a significant influence and I believe that my model over-estimates the amount of cabling used. The line of best fit to my data is logarithmic, so I’m guessing that my over-estimation gets worse the larger the wind farm is.
However, the number of onshore wind farms that have 30 or more turbines is low in the UK. I found 3 in this study of 30. Outside Scotland this is likely to remain the case due to constraints on space, and wind farms of 5 to 10 turbines currently typical in England and Wales.
Total copper use (without grid connection, because that is site specific) for onshore wind farms in the UK is 5.6 t/MW according to my model, with its warning attached for over estimation. A ‘safe’ estimate would be about 4 t/MW.
The methodology is more robust for offshore wind farms because the industry-standard cable topology seems to be single runs connecting rows of turbines with the single runs being gathered at the sub-station for conditioning and voltage step-up before transmission via the inter-connector.
Once you get to the inter-connector you are looking at a very serious piece of kit and a critical point of failure. You could loose one cable connecting a whole row a turbines and retain two thirds of output in a typical topology of a 3 row 30 turbine farm. Loose the inter-connector and you loose all output. Each inter-connector is a custom design. Fortunately, enough design detail is usually publicly available to make a good guess at the copper content (you need the total number of cables, number of turbines and their rated power output, and the transmission voltage to make an order of magnitude calculation). I only worked with 3-phase AC, since I could only find enough detail on one planned wind farm with HVDC.
So, to cut a long story short, my back-of-the-envelope calculation shows that offshore wind farms use copper at a rate of 9.6 t/MW by the time you connect them to the grid, with roughly one third of all copper being used in these installations contained in the inter-connector.
Not surprisingly the intensity of copper use goes up with length of inter-connector. My rough estimate is an additional 80kg of copper per MW per km of inter-connector. My R-squared is 0.31 on this, so there is a decent correlation, but its not brilliant. Note; distance to shore is not the same as inter-connector length. There may be a loose correlation between the two, but to give a couple of examples; one particular installation in German waters plans a 115km long inter-connector with 40km of that onshore, while another in the UK is 12km from shore but has a 43km cable route to the best available grid connection point.
To summarise;
Large wind turbines require around 2 tonnes of copper per MW of nameplate generating capacity.
Onshore they require a roughly similar amount in cabling infrastructure before you attach them to the grid, but my model overestimates the amount in large wind farms and this skews the result by a significant but undetermined amount. I estimate that my model doubles the cabling requirement in large onshore wind farms only, but I don’t have the hard data to back that up. A ‘safe’ estimate for total copper is therefore 4 t/MW onshore.
Offshore wind farms require more than double the copper per MW of installed capacity of their onshore cousins, but that includes the connection to the grid.
We will use 10 t/MW offshore and 4 t/MW onshore.
So this gives up our next estimate.
25GW offshore equals 250,000 tonnes of copper offshore
12GW onshore equals 48,000 tonnes of copper onshore
For a total of about 300,000 tonnes of new copper required just by UK wind power up to 2020.
Put that into context, in 2007 the UK exported 373,795 tonnes of copper and copper scrap (according to the BGS European Mineral Statistics)
Import trends from the same statistics show that copper will, effectively, no longer be imported into the UK as a raw material by 2011.
So even spread over 10 years at 30,000 tpa, this is a significant shift in raw material requirements for a country with virtually no manufacturing capacity left.
I’ll leave it there and discuss the potential implications in another post.
Random Energy Ideas Part Three
July 14, 2009
This is an idea that I first aired at the Falmouth Energy Week Conference in the innovation workshop, but that I’ll fully explain here.
There is a problem with new physical machinery in general. It doesn’t work very well.
There is a problem with innovation in the physical machinery of new low carbon energy systems. It costs a lot.
There is a problem with gaining people’s acceptance of new things in their landscape. We fear change and the unknown.
Here’s an idea to address all three issues a little at a time.
In order to increase the efficiency of the new physical machinery of a low carbon energy system it needs to be tested. It needs to be tested under different physical loads, under different environmental conditions and in different locations in order to find the optimum design characteristics such as low environmental impact, high output over as large a set of conditions as possible, low failure rates, etc.
But a lot of that can be done without building full-scale pilots and most of it can be done without building anything physical at all.
There is a common characteristic of many of the low carbon energy flux capture devices and indeed energy efficiency improvements i.e wind turbines, wave devices, tidal devices, high altitude kites, vehicle streamlining, etc, etc. That common characteristic is that their efficiency can be modelled using mathematical techniques called finite element analysis (FEA) and computational fluid dynamics (CFD). These models are difficult to design and take a lot of computing power to run, so only the larger companies and consultancies have their own in-house modelling capabilities.
How about we, the people, pay for a really good modelling tool to be built, but instead of forcing aspiring designers to buy it we give it away. But that’s not all. A key element of this issue is the ability to run lots of simulations to find the most efficient solution before building the damn thing, so we build the modelling tool to run on many, many computers at the same time using another technique called distributed computing. If you’ve seen SETI@Home or any of the other screensaver programs that take a chunk of data and use your home PC to analyse that data before sending the results back to a central point, you known what I mean.
And that’s the third issue dealt with too.
In making the general public part of the design process we automatically build in acceptance. We demystify and educate and empower through the simple act of involvement in the process. In contrast to the shared-ownership models of community acceptance that rely on a certain venality that rankles with many, this would be a genuinely altruistic action that has a side benefit of making sure that every participant had a vested interest, albeit a small and non-monetary one, in seeing these devices work. The designs would be publicly owned, the implementation would be privately done, and could include a portion of community participation if appropriate to the project.
If you wanted to you could have a split between device design and project implementation. If you think about a 3D terrain model (Ordnance Survey already has these) and placing turbines on it to minimise visual impact vs output, or even offshore for lines of sight. You could introduce a genetic algorithm to produce multiple design iterations, for example of a turbine blade and model the efficiency of each iteration to produce an efficiency envelope. You could model vibration modes at different operating speeds to minimise noise and failure rates. Every scenario that you could think of modeling would be accessible to every company.
There you go. Cheaper, community-embedded, device designs with a higher chance of acceptance and therefore a higher chance of implementation.
