Last Monday evening (8 February) I watched an online broadcast of a lecture entitled ‘Marine Renewables – Crossing the Valley of Death’, given at the Institution of Civil Engineers in London by Clare Lavelle, who is head of energy consulting in Scotland for Arup. Before that Clare worked for Pelamis Wave Power and before that for Scottish Power on its wave and tidal projects. Her presentation was very much a statement of the standard wave and tidal narrative that if only the government would spend enough money the technology would succeed. She even name-checked Professor Salter’s conspiracy theory and the one about how the UK lost its world lead in wind energy to the Danes.
Her view was that the only thing preventing these technologies climbing out of the valley is lack of funding, saying: ‘Pelamis didn’t fail because it didn’t work, it was just that the money ran out’. She suggested that government funding should be delivered via a combination of revenue support and, most importantly, capital grants for more full scale demonstration projects.
Roughly the same prescription, though cloaked in obfuscatory jargon, was given by the Offshore Renewable Energy Catapult in its November 2014 report ‘Financing solutions for wave and tidal energy’. This identified the problem as ‘… a lack of willingness from potential investors to invest in wave and tidal energy at present’, due to ‘lack of clarity on future investment return potential and timing’, and that ‘solutions to engage investors’ are needed. It goes on to say:
‘It is crucial that something is done to bolster the prospects of the industry in order to get investors back in play. Despite their reluctance to invest at the moment, the above mentioned investor groups will have a key role to play when their risk and return profiles start to align with the prospects for marine technologies and array projects.’
‘We have made some high level assumptions to estimate that, on its current trajectory, the tidal industry will need upwards of £100 million to get the first arrays of c.10MW to financial close. Additionally, we project that around £200 million of investment is needed to drive the wave industry along a path to commercial readiness.’
Though not explicitly stated, I think they mean that it will only be after the £300M has been spent that private investors' ‘risk and return profiles’ will begin to ‘align’.
DECC’s Marine Energy Programme Board also reached broadly the same conclusion in its February 2015 report ‘Wave and Tidal Energy in the UK – Capitalising on Capability’, calling for:
‘… funding at a suitable intensity to kick-start a transition to a commercial industry over the next 5 – 10 years.’
Finally, a paper has just been published by R. Bucher et. al. entitled ‘Creation of investor confidence: The top-level drivers for reaching maturity in marine energy’. This presents the results of a study into the views of a large number of wave and tidal industry stakeholders. It found, among other things, that the top three ‘drivers for success’ that the stakeholders themselves consider to be most important are:
- Government support
- Array-scale success
- Cost reduction
It seems, therefore, that there is an overwhelming consensus among marine energy stakeholders that:
- The technology is at an early commercial stage of development and is technically ready to build array-scale projects.
- Such projects are not getting built because the technologies’ costs are currently too high to attract private investors.
- The government, and only the government, can make it happen by funding early array projects until private investors gain enough confidence to start funding them themselves.
- Costs will fall due to the learning curve effect as capacity is rolled out at scale.
I think this is a fair summary of the ‘standard wave and tidal narrative’ that I alluded to in the first paragraph of this article. Unfortunately, it is wrong. To see why, we need to look at some facts.
In this article, the word ‘tidal’ refers to tidal current energy, not tidal lagoons, which are a different thing altogether.
According to EMEC’s web site, the following devices have been deployed at its full scale test sites:
|2005||AW Energy (mechanical testing only)|
|2009||Aquamarine Oyster 1|
|2010||Pelamis P2-001 (E.ON)|
|2010||TGL DeepGen 500kW|
|2010||Atlantis AK-1000 (subsequently replaced by the AR-1000)|
|2011||Aquamarine Oyster 800|
|2011||Andritz Hydro Hammerfest HS1000|
|2012||Pelamis P2-002 (ScottishPower Renewables)|
|2012||Alstom (formerly TGL) 1MW ‘DeepGen’ (replaced earlier 500kW version)|
|2013||Voith Hy-Tide 1MW|
As well as these, there have also been:
|2002 & 2003||The Engineering Business||Stingray||Yell Sound, Shetland||Not grid connected|
|2002||Marine Current Turbines (MCT)||Seaflow||Lynmouth, Devon||Not grid connected|
|2008||Sea Generation Ltd (MCT)||Seagen||Strangford Narrows,
|2011||Ocean Power Technologies||Power Buoy PB150||Cromarty Firth||Deployed for 6 months. I don’t think it was grid connected.|
Ofgem publishes data for Renewables Obligation Certificates (ROCs)—tradable green certificates that until 2017 are the main way that renewables are subsidised in the UK—and ‘Renewable Energy Guarantees of Origin’ (REGOs)—certificates that are used by electricity suppliers for Fuel Mix Disclosure under EU law and which have no monetary value.
To claim ROCs and/or REGOs you first need to get your generating station accredited by Ofgem. This is to ensure that it is indeed renewable and therefore eligible. Ofgem has so far accredited ten marine energy generating stations. These are:
|Generating Station||Capacity (kW)||Organisation||Status||Notes|
|Eday Berth 4||200||OpenHydro Group Ltd||Live|
|SeaGen||1200||Sea Generation Limited |
|Vagr Atferth||735||E.ON UK||Live||Pelamis P2|
|Billia Croo Berth 6||800||Aquamarine Power Limited||Live||Oyster|
|CLADDACH FARM||147||Voith Hydro Wavegen Limited||Live||Limpet|
|brummer||1000||Voith Hydro Ocean Current |
Technologies GmbH & Co. KG
|Orcadian Wave||711||ScottishPower Renewables |
|Eday Berth 2||1000||Tidal Generation Limited||Live|
|Eday Berth 1||1000||ANDRITZ HYDRO Hammerfest |
|Ness of Quoys||5700||MeyGen Limited||Preliminary|
Ocean Power Technologies, Scotrenewables, Seatricity, Wello Oy and the Atlantis AK-1000 / AR-1000 appear not to have any accredited stations.
Table 4 below shows the total number of megawatt-hours (MWh) in respect of which each type of certificate was issued. These should be the same but, for some reason that I don’t understand, they are slightly different. They are, however, close enough that they tell the same story.
|Eday Berth 2||1364.6||1452.85|
|Eday Berth 1||1056.6||1160|
|Billia Croo Berth 6||0||12|
The discrepancy between ROCs and REGOs is particularly large in the case of SeaGen. Figures 1 and 2 below show the same data as a function of time between April 2006 and February 2015. This shows that Seagen appears to have received ROCs but not REGOs for a big chunk of time between 2009 and 2011. Apart from that, the two sets of data are very similar.
One of the stations, Eday Berth 4, doesn’t appear in the data at all and another, Billia Croo Berth 6, produced 12MWh’s worth of REGOs but no ROCs. OpenHydro announced in 2008 that it had delivered electricity to the grid, but doesn’t appear to have received any ROCs or REGOs. Maybe it was less than 1MWh? Aquamarine Power states that both Oyster 1 and Oyster 800 delivered power to the grid, but there is similarly little evidence of it in the data.
Figures 1 and 2 show the number of MWh in each Output Period, which is always one calendar month. This means that the y-axis (i.e. the height of the bars) can be interpreted as power output in units of MWh/month. The graphs only go up to the end of February 2015 though I think Ofgem’s policy is to release data after a time lag of three months. This would imply that there has been no generation between March and November 2015 inclusive. It is possible, but unlikely, that in the three months from the beginning of December 2015 to the end of February 2016 there have been significant developments that we don’t know about yet.
To judge how good or bad these data are, we first need to know how much energy each of the devices was expected to produce.
First, the bad news
SeaGen is clearly the most successful of the devices shown in Figures 1 and 2 above. It was installed in Strangford Narrows, Northern Ireland in 2008. Its peak output was in August 2010. During that month it generated 522MWh. I am going to assume that this represents its output at 100% availability. That is, I reckon that in August 2010 MCT managed to keep SeaGen spinning continuously for the whole month without stopping. So its output corresponding to 100% availability is 522MWh/month.
In Figures 1 and 2 there are two distinct periods of SeaGen activity. The first is between April 2009 and July 2011. The second is between December 2011 and December 2013 inclusive, a period of 25 months. During that period it generated 6496.5MWh (according to ROCs) or 6519MWh (according to REGOs). Dividing the larger of the two by the product of 522 and 25 results in an average availability of almost exactly 50%.
For any other kind of power station 50% availability would be regarded as rubbish, but for a marine energy device it’s pretty damn good and represents a major achievement. Of course, we don’t know how hard it was to achieve—whether they were able to leave it alone to do its stuff or whether they had to tinker with it continually to keep it going. Actually, it must be the latter because if they were able to do the former it would have have achieved 100% availability.
However, it wasn’t part of the plan, or even foreseen, that only such a low level of availability would be reached. According to Douglas, Harrison and Chick (2008), which was published before SeaGen was installed, the ‘Technical specifications of Seagen Marine Current Turbine’ included ‘Reliability > 90%’. But that’s OK because SeaGen, like SeaFlow before it, has been retrospectively designated ‘a research platform’.
GE’s (formerly Alstom’s, formerly Rolls Royce’s, formerly TGL’s) device and Andritz Hydro Hammerfest’s (AHH’s) device are both located at EMEC’s tidal test site at the Fall of Warness near the island of Eday. To estimate their monthly energy output corresponding to 100% availability I’m going to use a quick back-of-the-envelope type of calculation that assumes that the current speed is a simple bi-sinusoidal function. This requires only two parameters to characterise the tidal current regime; mean spring peak current speed, as, and the mean neap peak current speed, an. Lawrence, Kofoed-Hansen and Chevalier (2009) present the results of measurement and modelling for a number of locations around EMEC. Visual inspection of the graph in their Figure 16 (location A04) suggests that good estimates for the two current speeds would be as = 3.1m/s and an = 2.0m/s.
GE’s device has a nameplate capacity of 1MW and a rotor diameter 18m and AHH’s device also has a nameplate capacity of 1MW but has a rotor diameter 21m. I’m going to arbitrarily assume that their ‘coefficient of performance’, Cp = 0.45 and that they both have a cut-in speed of 1 m/s. With these assumptions, the power output of a 1MW turbine with a rotor diameter of 21m (red line) and a 1MW turbine with a rotor diameter of 18m (blue line) would look like the curves in Figure 3 below, which only shows a few hours so that you can see the shape of the peaks.
Integrating these curves over a whole year and dividing by 12 gives the average energy output that would be expected in a month if the turbines suffered no downtime. For the 18m diameter turbine this comes out at 285MWh, and for the 21m diameter turbine it is 340MWh. Because of the cube-law relationship between power output and current speed these results are extremely sensitive to errors in the assumed values of as and an, and so should be regarded as only very rough.
If they were 100% available, i.e. suffered no downtime, we might, therefore, reasonably expect that GE’s turbine would produce something in the region of 285MWh in a month and AHH’s would produce something in the region of 340MWh.
Eday berth 1 (AHH) had a burst of output between September 2013 and March 2014 inclusive, a period of 7 months, during which it delivered 931MWh (according to ROCs) or 1003MWh (according to REGOs). Dividing the larger of the two by the product of 340MWh/month and 7 months gives an average availability of 42%.
Eday berth 2 (GE) had a solid burst of output between April 2014 and February 2015 inclusive, a period of 10 months, during which it delivered 1030MWh (according to ROCs) or 1066MWh (according to REGOs). Dividing the larger of the two by the product of 285MWh/month and 10 months gives an average availability of 37%.
As I remarked above, these estimates are very rough and should be regarded as having a wide margin of error.
It would be reasonable to conclude from all this that tidal-current energy technology is in a no-man’s-land between success and failure. It has performed well enough not to be regarded as having unambiguously failed, but not well enough to be regarded as ‘proven’ and so to justify building large scale commercial projects. In addition to being low, the above estimates of availability depend on the judicious choice of time period over which they were calculated. While this period was two years for SeaGen, it was only ten months for GE’s turbine and seven months for AHH’s.
There is, of course, an array under construction at the moment, the Meygen project, which is being built in one of the scariest bits of water in the world. I hope they succeed, but the odds don’t look good.
Now, the worse news
Unlike tidal, wave has no simple one-line formulae that can be used to make a quick back-of-the-envelope calculation of how much energy a device should produce, short of full-blown hydrodynamic modelling. We can, however, use Pelamis Wave Power’s famous power matrix, which can be found in the report Pelamis WEC – Conclusion Of Primary R&D, as well as in a series of brochures that PWP produced over the years but which don’t appear to be on the web any more.
Grid square id 4246 in the wave dataset of DECC’s Atlas of UK Marine Renewable Energy includes EMEC’s wave test site off Billia Croo on the west coast of Orkney Mainland. This contains the following data:
The atlas gives the parameters Hs and p. The values of Te in the last row of the table were calculated by me using the formula Te = p / (0.49Hs2). Feeding these numbers into the power matrix gives the following figures for the power output of a device:
It is interesting that the power output values in the above table are much smaller than the device’s nameplate capacity of 750kW. This is because the machine to which the power matrix refers was designed for more energetic locations farther from shore. The conditions at the EMEC test site correspond to the blue shaded area near the top right of the power matrix. I don’t know whether the machines that were actually deployed at EMEC were so designed, or whether they were optimised for the location where they were actually deployed, in which case they may have been capable of producing more. If that was the case, however, the actual performance would be worse by comparison.
Figure 5 below compares this with the energy actually generated by the two Pelamis P2 devices, E.ON’s Vagr Atferth and Scottish Power’s Orcadian Wave. I have presented only the REGO chart because the RO chart is almost identical and wouldn’t add anything to the discussion. This chart speaks for itself and doesn’t need an availability calculation.
Investors reluctant to invest in crap technology shock
For both wave and tidal devices, it is not possible, based on the performance of the devices that have been built so far, to predict how much energy the next project will generate, and predictability is what investors require before anything else.
People will say, inevitably, that the above projects are actually great successes because they have managed to generate anything at all, or because they have ‘survived’, in the sense of not having been destroyed, a winter or two in the water. Ten GWh, or even one, can sound a lot if you say it slowly and stress the capital ‘G’. It is, however, clear that neither wave nor tidal has performed well enough to be regarded as proven.
What was predictable was that no one would win the Saltire Prize, though the exercise may not have been entirely without its uses.
The nature of the problem
The only reason why output like that shown in Figures 1 and 2 would be produced is that unexpected things have happened. One or more of the dreaded Rumsfeldian ‘unknown unknowns’ have revealed themselves. These could take two forms; the device’s capture efficiency might turn out to be less than predicted or it might be very unreliable, or a combination of the two. I think it’s reasonable to discount the first of these, especially for tidal, because the theory of their operation has been extensively studied for decades at some of the world’s top universities. I don’t doubt that a Pelamis or an Oyster will produce what its designers say it will produce, during the periods that it is producing. This leaves the other alternative, that the devices have not operated reliably.
Another explanation, that I expect some people will put forward, is that some device developers, especially the ones that haven’t been grid connected, have actually performed well but have chosen not to claim ROCs so as to keep commercially sensitive information secret. It is, however, inconceivable that anyone would do that. Firstly, they wouldn’t be able to afford to forgo the ROC income which, if they really had performed well, would be large. Secondly, they would also have to forgo brown electricity sales, because any supplier that bought their electricity would insist on having the REGOs. Finally, if they really had performed well they wouldn’t be trying to keep it secret, they would be shouting it from the rooftops.
It is frequently said that the main challenge that wave and tidal face is cost reduction, but it is clear from the above discussion that the technology actually faces two challenges, reliability and cost reduction, in that order. There’s no point in trying to engineer the cost down if you can’t keep the machine going for more than few days at a time. Downtime is a double whammy because you get no revenue when the machine is down and it costs money to get it going again. To stand any chance of operating commercially the technology must achieve predictable, ideally high, levels of reliability.
Although people have earned a living from the sea for thousands of years, they have always avoided big waves and strong currents. The only permanent structures that have been built in the sea, oil platforms and, more recently, offshore wind turbines, similarly avoid areas where there are strong currents, and the parts of the structure that pass through the splash zone present a fixed, rigid, hard, smooth surface that reflects waves back into the sea, as do the hulls of ships. Machines that actively absorb the energy contained in waves and currents and channel it through a complex mechanism are asking for trouble.
Thies, Flinn and Smith (2009) and Delorm, Zappalà and Tavner (2012) have carried out probabilistic reliability modelling on various marine energy devices. Their work was hampered by a lack of available data, which they attributed to the secretive and competitive nature of the industry, but using surrogate data from elsewhere found that marine devices are likely to suffer lower reliability than offshore wind turbines.
Marine energy devices, however, actually need to be more reliable than offshore wind because the cost of accessing the devices is likely to be higher.
While it is possible to ‘manage’ reliability issues through well designed O&M strategies and component redundancy, this can only get you so far. To actually solve the problem it will be necessary to design out the unreliability at a component level.
None of the companies developing marine energy devices expected their development to have taken so long or to have cost so much. For example, Figure 7 below shows an extract from a presentation by Richard Yemm at an IMechE seminar in 1999:
This is how technology development is supposed to happen. First an R&D phase in which fundamental questions are answered, then a demonstration phase in which the technology is demonstrated to perform as expected with no surprises, or at least only a few small ones. I’m sure all the other device developers had the same expectation when they started. It’s a pity they haven’t turned out that way. As Oscar Wilde would have said, to find unknown unknowns in your first full scale demonstration is unfortunate, to continue finding them after you have built six looks like carelessness.
If the demonstration phase turns out not to demonstrate what it was intended to demonstrate, then the solution is not to build more and more of them until eventually one of them works, because it probably won’t. There’s an old saying to the effect that if you carry on doing what you’ve always done, you’ll carry on getting what you’ve always got, which in this case is a lot of big machines that don’t work. The development process should not involve a long sequence of full scale devices, one after the other, each making only a marginal improvement over the previous one.
RenewableUK’s report Capitalising on Capability claims that in the wave and tidal sector there has been ‘nearly £450 million spent to date in the UK supply chain’. This means that the actual amount of money that has been spent must be larger than this. In an interview in The Herald Richard Yemm said:
‘… of the £95 million invested over the past 17 years … the net contribution of Pelamis to the Scottish economy to date is between £65m-£70m positive.’
Applying the ratio 95/70 to RenewableUK’s figure gives £610M. This is more than twice the £300M that the Offshore Renewable Energy Catapult says is needed to get the sectors to commercial reality. I don’t know about you, but to me it doesn’t look like the sector is two thirds of the way to commercial reality. This suggests that the Catapult’s figure may be an underestimate and that the answer is not for the technologies to continue on their ‘current trajectory’.
My suggestion is that instead of spending over £300M on more failed full-scale ‘demonstrations’, the money should be redirected to focus on solving the reliability problem. This would most likely involve research to advance the science and technology of how to make ultra high reliability components and machines. This could include, for example, the development of new:
- surface treatments
- design methods
We need research to better understand the physical processes that are the root causes of these failures and to develop ways of slowing them down or stopping them. In most categories of technology reliability has evolved to a level where the benefits of further improvement are not large enough to justify the effort of achieving that improvement. For marine energy, this level is different.
Such an approach may have benefits in a wide range of other areas of technology too, and a programme along these lines need not be limited to marine energy.
Wave and tidal energy technologies are like Leonardo da Vinci’s flying machines—the ideas are good but they may have to wait for developments in unrelated areas of science and technology before they can be made to work.
Of course, when this problem has been solved, there’ll still be the issue of cost reduction but, when the technology is at a stage that it ready to tackle that issue, the market conditions may be different.
© Copyright 2016 Howard J. Rudd all rights reserved.