Wednesday, 11 November 2015

EWTEC 2015 - a leisurely sail through the papers

There were so many talks in this year’s EWTEC that I found interesting, it was impossible to pick just a few. Here are highlights that I think will be of general interest.

Fathoming out the true bearing

Several papers discussed the incentives and development route required for commercialisation of wave power.

Weber and Laird’s paper described the early stages of a project on structured innovation of wave energy converters (WECs), based at two US laboratories, NREL and Sandia. The project plans to use several best-practice structured innovation techniques, including functional decomposition and theory of inventive problem solving (TRIZ – a Russian acronym), to generate new WEC concepts. The plan is to validate these concepts with tank tests, and make the optimised versions available to industry for commercialisation. This development route provides a way around the constraints faced by industry due to intellectual property restrictions. Work so far has been on the refinement of the TPL (Technology Performance Level) metric for use in this project, and specification of the functional requirements of WECs. There are planned workshops in Europe and the US to allow input from industry to the TPL metric and functional requirements.

Bucher and Jeffrey interviewed 44 stakeholders on the challenges facing marine renewables. The two top challenges were also the two things seen as most benefiting the industry, namely ‘technology learning’, and ‘marine operations experience’. Pilot arrays would deliver this, but the two top-ranked risks for arrays were ‘achieving funding’ and ‘uncertain performance’. These are linked, as funding generally requires certainty about performance – and vice versa. The recommended route to this crucial intermediate milestone of first array success is to create motivation for collaboration between natural competitors, as well as ‘incubation rooms’ for immature technology. The paper also made a poignant distinction between two types of complexity. ‘Detail complexity’ is undesirable, and can be reduced for example by applying systems engineering and standardisation. ‘Dynamic complexity’ cannot be reduced, and characterises large-scale engineering projects. What is missing in marine engineering is an understanding of the inherent dynamic complexity in commercialising wave energy, and how to manage it.

The recommendation of this paper was backed up by the ‘Danish partnership for wave energy’. Nielsen reported that their attempts at competitive collaboration have been very fruitful. They also announced plans for collaboration on sub-systems. There are plans for test rigs to trial different types of PTO side by side, as well as different materials and components. They also recommend a series of publicly supported demonstration parks, with increasing capacity, distance from shore, and an associated decreasing production tariff.

Another couple of Danes, Andersen and Frigaard, questioned whether conventional support mechanisms could incentivise continuous improvement. Production tariffs incentivise continued production, but risk going over budget if uncapped. Capital grants do not risk going over budget, but do not guarantee project continuation. Both struggle with public acceptance: no one wants to pay for a prototype that promptly breaks, or pay over the odds for electricity. The difficulty is that neither of these options truly incentivise what we want, which is technology learning. Andersen and Frigaard suggest incentive-based capital support. This is like a capital grant spread out over the demonstration project lifetime (e.g. 5-10 years). Portions of funding only become unlocked when targets are met. Examples of targets are electrical or mechanical power capture for a given resource. The targets become increasingly tougher as the project progresses. They presented a case study for a hypothetical WEC with the rather pleasing acronym ‘WHOOPS’(!).

An anchorage where we batten down the hatches

There were several presentations that questioned whether Scotland and Ireland were the best places for first arrays. Surprisingly, the presenters were based in Scotland and Ireland. I predict collaborations with Chile and Mexico in the near future!

Pascal et al.’s paper presents new ways of processing the wave resource scatter diagram that are so intuitive and insightful, the only reasonable response is a slap to the forehead with the caption ‘Why I didn’t I think of that before?’. The paper uses only the scatter plot data: i.e. occurrence of sea states at a given site in terms of \(H_s\) and \(T_e\) (height and period derived from the spectrum). Three new ways of presenting this data are offered:
  1. Histograms of both occurrence and energy contribution for each period. This can be used to show how well occurrence and energy contribution are matched. This indicates the amount of design compromise required in the choice of the WEC’s frequency response.
  2. Cumulative distribution of both occurrence and energy against \(H_s\). This shows what wave height a WEC needs to operate in to interact with a given percentage of the annual energy, and what percentage of time it will be extracting power.
  3. Cumulative energy distribution against sea state power (per metre wave crest). This shows the power of the sea states a WEC needs to operate in to interact with a given percentage of the annual energy. The sea states corresponding to 90% and 95% of the annual energy are marked. The ratio of these to the site’s mean power is a mean:maximum ratio that applies specifically to the operating envelope (assuming a storm weathering mode kicks in at either the 90% or 95% sea state) . This graph can be used to investigate the PTO rating, operating envelope, and capacity factor.
These plots were applied to four locations: EMEC (Scotland), Belmullet (Ireland), Pt Arguello (California), and Pilot Zone (Portugal). Of these, EMEC proved the most challenging for WEC operation. It had the greatest ratio between mean power and the maximum operating sea state power. For EMEC and Belmullet, occurrence and energy distribution over \(T_e\) are mismatched, indicating WECs with broad bandwidths and low capacity factors are required. Both require operation in high \(H_s\) to interact with 90% of the wave energy: 7m / 8.5m (EMEC / Belmullet). The Californian site faired the best overall: the occurrence and energy histograms were closely matched, operation in waves up to 4.5m would capture 90% of energy, the mean power is higher than EMEC’s, yet the sea state required to interact with 90% of the annual energy is around half of that for EMEC.

The paper also considered sea states that were common to all four sites. Results showed that seas most common at EMEC were not common at other sites. e.g. sea states that together represent 70% of occurrences at EMEC only represent 32% of occurrences at Belmullet. This highlights the limitations of EMEC for estimating power matrices relevant to other sites.

de Andrés et al. considered the timing of high energy sea-states, i.e. the distribution of weather windows. This impacts availability and operating expenditure. OpEx per kWh was used to compare sites around the world. Assuming that vessel waiting times bore a daily cost, the British Isles had the highest indications of maintenance costs, while Mexico, Nova Scotia, Brazil, Japan and New Zealand had sites which looked promising for maintenance. New technologies need accessible sites to keep down the maintenance costs. Accessibility is less important for highly reliable mature technologies. This suggests that the British Isles are not the best location for prototypes, even if they might be suited to more reliable devices. There exists some uncertainty, because if waiting time costs are limited to 15 days (realistic market arrangement) then Ireland jumps from the worst to the best category for maintenance costs. Note that the annual energy estimation assumes a constant capture width, i.e. no rated power. This assumption would cause OpEx per kWh to be underestimated, particularly for sites where large mean:extreme ratios will lead to high capacity factors.

This paper also suggested a new and useful way of defining availability: it takes into account the percentage of resource power not captured due to a failure. ‘Resource availability’ is always less than ‘time availability’. Note again that this assumes the operating envelope extends up to the highest seastates, i.e. there is no storm mode.

Ringwood and Brandle’s paper considers the time series wave elevation data rather than the usual spectral parameters. Several sites across the world are summarised on an interactive Google map. For each location, the coefficient of variation (standard deviation / mean) is given. Plots show how both the resource and the variation fluctuate annually and monthly. The diagram below showing sites in terms of mean resource and variability could be used to guess which locations may have been considered by Pascal et al. but not included in their paper for commercial reasons! For example, the Chilean resource is excellent in terms of variability, and has sites with a range of mean powers. In contrast, the Irish and Scottish sites considered had relatively high annual and monthly variability. During the audience questions, I asked Prof Ringwood about the implications for UK and Irish test sites. He answered that wave energy’s success depended on the choice of the most economically favourable test sites.

In Prof Ringwood’s talk, he made a strong case for variability being a cost-driver. He argued that it was straight forward to operate at, and optimise a design for, a specific operating point. However, there are costs to adapting to a variety of operating conditions, in terms of design complexity, capital cost, and performance compromise. Control systems are required to extend the operating conditions.

Characteristics of different wave resources; replotted data from the Ringwood and Brandle paper.

Tackling the mean:extreme ratio (one can go overboard on nautical puns)

There was a common theme to the three papers on resource just described: the recognition that reducing the mean:extreme ratio was key to making wave energy economic. There were a few papers that extended this theme to the design of wave energy devices themselves.

Murtagh and Walsh addressed the question ‘What type of PTO would contribute to cost effectiveness and robustness?’ They compared the most common strategy: optimal linear damping (force proportional to velocity), to a strategy that resulted in lower capture efficiency, but a better mean:extreme ratio: coulomb damping (constant force). Tank tests at Queen’s University were used to compare the strategies. As expected, Coulomb damping resulted in lower forces, higher velocities, and higher strokes. Lower forces mean that lower rated PTOs can be used. While higher stroke could be a problem for linear PTOs with end-stop limits, it would allow size reduction in rotary PTOs.

A parametric cost study using guidance from the wind industry estimated that the generator required for coulomb damping would be under half the price of a conventional generator using linear damping. This is a significant saving, as the Carbon Trust estimates that the PTO system could account for half the project capital expenditure. The tank tests suggested that coulomb damping resulted in 20% less power capture. However, the authors referred to a paper by Falcao that suggested the two strategies resulted in similar net power production.

Chapman et al. described tank testing of the WaveSub device. It consists of a small spherical body positioned above a larger reaction plate, both of which are submerged. Power is extracted from the relative motion via four tensioned cables. The survival strategy is to increase the submergence during storms, thereby limiting extreme loads. Data from Belmullet was used to simulate storm waves in the Plymouth tank. Results were compared for the normal operating position and the survival configuration. Rainflow analysis was conducted on measured cable loads. In the survival condition, peak loads were not reduced, as the pre-load tension was higher. However, the amplitudes of the variation in load were significantly reduced. This has benefits in terms of fatigue and shock loading, and the clearer design load case presented by the reduced load range.

Tom et al. presented a design of pitching flap that has a controllable hydrodynamic surface. It can incrementally adjust the area that is excited by the waves by using louvered flaps. This has potential to decrease the peak loads, and hence lower the mean:extreme ratio and increase the capacity factor.

Hals Todalshaug et al. wrote about CorPower; a concept that gives serious consideration to increasing the ratio of revenues to costs. It does this in two ways. In common with other devices like SeaBased, a surface-piercing buoy is pulled down with tensioned cables. Along with the usual advantages of a pretension system, namely transmitting forces in tension rather than compression, capturing power from both heave and surge, and adjusting for tidal variation, CorPower uses pretension to increase the wetted volume for a given mass. Effectively, some of the mass costs are replaced with pretension costs.

The second mechanism improves the mean to extreme ratio via the ratio of operational to survival loads. The basic idea is not new: a low-mass heaving buoy has a low natural period, so will not attract high loads in failsafe and storm survival modes. During operation, phase control is used to raise the effective natural period to match the periods of economically viable seastates. The use of simple proportional-integral control allows this to be done using real-time measurements of displacement and velocity (so no prediction of future waves is required). Up till now, the only way of implementing this has been to use the PTO equipment to provide ‘negative spring’: a force proportional to displacement, in the opposite direction to the restoring force. An economic solution has not yet been found –mainly because it requires an over-sized generator that runs mostly at part load, which is not good news for the ratio of mean to extremes.

The exciting bit about this paper is the WaveSpring: a separate component that provides the negative spring. This took me by surprise, because I was not aware such a thing existed, having previously heard of this being discussed in the same context as a ‘skyhook’. However, negative springs are already in use in other engineering applications. If the losses and costs of the WaveSpring are less than those associated with providing reactive control using an oversized generator, including associated gear-box and power electronics, then this is a concept worth getting excited about. During the audience questions, Yukio Kamizuru asked about the heat losses in the pneumatics of the WaveSpring (they were neglected in the paper). Jørgen Hals-Todalshaug answered that thermal losses were recognised and said full-scale projections were promising. Many of the results in this paper are available in the associated Marinet report.

Chocoblock with information

There were several papers that were notable for their contribution to improving the collective understanding of the wave energy challenge.

Combourieu et al. presented results from phase I of the Wave Energy Converter Code Comparison project. They compared four codes for simulating wave energy converters: InWave, WaveDyn, ProteusDS and WEC-Sim v1. The same device was modelled: a reference case based on the Langlee device. While hydrodynamic coefficients were very similar, the drag coefficients and pitch responses were surprisingly different. There is no way of knowing which is correct, apart from waiting until Phase II, which will compare the codes with experimental data.

Folley et al. discussed the differences between extracting power from the vertical versus horizontal water particle motions in waves. The main thrust of Dr. Folley’s talk was that people’s particular experience of wave energy largely determined their perceptions of the challenges faced, and hence the solutions required. He gave the example of ‘luddites’ vs ‘control freaks’, i.e. the ongoing debate about the merits of control. People who work with heaving devices tend to be pro-control, because control is required to make heaving devices respond well at wave periods of interest. Pitching and surging devices however do not require control to have good response at higher wave periods, so people working on these devices tend to fall into the ‘luddite’ category. This paper is however more than just a plea for improved mutual understanding; the discussion about the differences between heave and surge is an excellent piece of science communication. It gives an intuitive feeling how the modes of motion impacts the hydrodynamic response.

Adams et al. describe best practice for uncertainty calculations of energy yield for marine renewables. The paper is a summary of a Catapult study involving industry-wide stakeholder interviews: 'Wave and Tidal Energy Yield Uncertainty - Reference Document (July 2015)'.  Types of uncertainty are broken down according to a taxonomy developed by the Catapult, which can be found in table 1 of the literature review associated with this study: Wave and Tidal Energy Yield Uncertainty - Literature Review (July 2015).

Also worth mention are a paper by Retes et al. describing non-linear terms that may be included in a model of a wave energy converter, and a paper by Davidson et al. which gives a practical guide to using OpenFOAM, an open source CFD code.

Tales of the high seas from salty sea dogs

When it comes to improving collective knowledge, the experience of developers is vital to ensure academia knows what problems to tackle. Yet sharing this experience could benefit competitors or alarm investors, so lessons learnt are rarely shared. Yet a few developers did share lessons this EWTEC. Aquamarine’s openness is of great benefit to the community, but the context is clearly difficult for the people involved. The main message in the paper by O’Boyle et al., that full scale trials deserve continued funding, is poignant given Aquamarine’s recent troubles.

The O’Boyle paper described lessons learnt while testing the Oyster 800. These included:
  • Having separate control and data acquisition systems rather than an integrated SCADA was valuable as it allowed closed-loop control.
  • The ability to cable the wave measurement devices (AWACs) directly to shore gave pseudo-real-time access to wave data, and eliminated problems with time-stamp synchronisation and undetected faults.
  • It is difficult to design a control system based on numerical simulations, which won’t include all constraints and external influences. Modifications are required during commissioning. To spot bugs, data should be collected during commissioning and compared to a simulation. Sources of discrepancy include the precise component properties and the control system parameters. With hundreds of such parameters, it is essential to identify which key measurements can be used diagnostically. For example, a discrepancy between the simulation and prototype demand signal identified a bug that robbed 15% of the power. The fix involved adjusting some parameter values in the prototype’s control system.
  • There was no direct measurement of friction in the hinges, seals and bearings, which accounts for typically ~10% of the flap torque. This was addressed by adding uncertainty bounds to reported data.
  • Accurate wave measurements reduce the uncertainty in capture width calculations and model-prototype correlation studies. The initial positioning of the AWACs gave data with high uncertainties. The AWACs were too far from, and at different depths to, the prototype. Both were relocated closer to the Oyster to improve the data quality.
  • Mean wave direction is an important reading from the AWAC. It was found that the internal compass which gave the reference direction was distorted by the presence of the AWAC mounting and the prototype. This was addressed by doing a gyro-compass survey of the installed AWACs.
  • The AWAC also gives the tidal level. Estimations were adjusted using measurements of atmospheric pressure.
  • Numerical models were shown to be better than tank tests for investigating full-scale results. Research tools should not be validated only on power capture but also on dynamic similarity.

Another Aquamarine paper, by Lamont-Kane et al., describes experimental investigation of slamming events in the Oyster. Preliminary results suggest the most severe slams occur in steep seas with short periods, rather than the most energetic sea states. The study used pressure sensors and underwater images to explain different slamming cases in terms of how the flap was moving relative to the water particles.

It is interesting to compare Aquamarine’s model-prototype correlation study with the one performed by DNV for WaveRoller. Child et al. gave a detailed description of the numerical model used. WAMIT gave the usual hydrodynamic coefficients, as well as pressure variation over the flap, which was converted to strains at positions where the prototype had strain gauges installed. Excitation forces were calculated for a series of flap angles, while radiation coefficients were fully linear. Response was calculated using WaveDyn, which included non-linearities such as drag and PTO torque. The time series wave data was not available from the sea trials, so comparisons between model and prototype were statistical. Simulation results were well matched to the prototype in moderate seas, and less well matched in energetic seas.

Carnegie presented key results from an internal review of their sea trials at Garden Island, Australia. Fiévez and Sawyer acknowledged that the advantage of redundancy had been underestimated. Staggered installation of several prototypes turned out to be useful in several ways. Faulty components were only discovered after the installation of the first prototype. These faults were fixed on the next two before they left the factory. Instrument failure on the first prototype was fixed quickly by robbing components from the devices still in the factory.

Lessons learnt also included management issues. Fiévez and Sawyer argued that subcontracting out the installation hinders the design process, as installation and recovery must be designed in parallel with the device. They also found that supplier’s terms and conditions were often self-serving.


My attendance at EWTEC and therefore my ability to report back on it was very generously sponsored by Jochem Weber. Many thanks to the authors who proof-read my summaries of their papers, and to Prof Ringwood and Prof Clement for the images they contributed.  


08D5-1 J. Weber and D. Laird, “Structured innovation of high performance wave energy converter technology” in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

09C1-4: J.C. McNatt, D.I.M. Forehand, G.S. Payne and V. Venugopal, “Experimental analysis of cylindrical wave fields” in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

10A2-3: N. Tom, M. Lawson, Y. Yu, and A. Wright, “Preliminary analysis of an oscillating surge wave energy converter with controlled geometry” in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

10A1-1: B. Child, J. Roadnight, and C. Ridgewell, “Validation of a numerical method to predict the structural response of a wave energy converter against at-sea data” in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

08C1-3: M.P. Retes, G. Giorgi, and J. V. Ringwood, “A review of non-linear approaches for wave energy converter modelling” in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

09A1-3: P. Lamont-Kane, A. McKinley, A. Henry, J. Nicholson, M. Folley, and B. Elsäßer, “Estimating extreme loads on an oscillating wave surge converter” in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

07D4-1: R. Pascal, A.T. Molina, and A.G. Andreu, “Going further than the scatter diagram: tools for analysing the wave resource and classifying sites” in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

10B4-4: J.V. Ringwood and G. Brandle, “A new world map for wave power with a focus on variability”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

09C5-3: A.D. de Andrés, H. Jeffrey and R. Guanche, “Finding locations for wave energy development as a function of reliability metrics”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

07D4-5: E. Mackay “A unified model for unimodal and bimodal wave spectra”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

09B5-1: R. Bucher and H. Jeffrey, “The strategic objective of competitive collaboration: managing the solid market launch of marine energy”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

09B5-3: K. Nielsen, J. Krogh, H.J. Brodersen, P.R. Steenstrup, H. Pilgaard, L. Marquis, E. Friis-Madsen, J.P. Kofoed, “Roadmaps for the development of technologies related to Danish wave power systems”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

09B5-5 : M.T. Andersen and P.B.Frigaard, “Incentive-based financial support scheme for immature renewable energy systems”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

09C5-1: N. Adams, S. Livermore, J. Sinfiled, V. Coy and R. Torr, “Developing best practice in uncertainty assessment for wave and tidal energy projects”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

09B3-4: C. Murtagh and P. Walsh, “Wave energy conversion: linear vs coulomb PTO daming strategies”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

08A3-1: J. Chapman, D. Pérez-Torres, A. Baldini, J-B Le Dreft, I. Masters, G. Foster, G. Stockman, D. Greaves, and G. Iglesias, “Validation of storm load limitation of a novel wave energy converter using scale model testing”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

09B3-2: J. Hals Todalshaug, G. S. Ásgeirsson, E. Hjálmarsson, J. Maillet, P. Möller, P. Pires, M. Guérinel, M. Lopes, “Tank testing of an inherently phase controlled wave energy converter”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

07D1-3: A. Combourieu, M. Lawson, A. Babarit, K. Ruehl, A. Roy, R. Costello, P. Laporte Weywada, H. Bailey, “WEC3: Wave energy converter code comparison project” in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

08B1-1: M. Folley, A. Henry, and T. Whittaker, “Contrasting the hydrodynamics of heaving and surging wave energy converters”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

09B1-1: J. Davidson, M. Cathelain, L. Guillemet, T. Le Huec, and J. Ringwood, “Implementation of an OpenFOAM numerical wave tank for wave energy experiments”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

08D1-1: L. O’Boyle, K. Doherty, J. van ‘t Hoff, J. Skelton “The value of full scale prototype data – testing Oyster 800 at EMEC, Orkney”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

0D1-4: J. Fiévez and T. Sawyer, “Lessons learnt from building and operating a grid-connected wave energy plant”, in Proceedings of the 11th European Wave and Tidal Energy Conference, Nantes, France, 2015

 Image credits

The photo of the EWTEC regatta, copyright EWTEC, was kindly provided by Prof Clement.
The graph from the Ringwood and Brandle paper was kindly replotted by Prof Ringwood. 


  1. In case you have not seent it, a numerical treatment on the effects of a 'snap-through' negative spring mechanism on a WEC is available here:

    1. Thanks for the links Adi - the work of these researchers is new to me.

      The basic dynamics are the same as the WaveSpring. The implementation is not specified - any type of spring could be used. With the WaveSpring on the other hand, the implementation is quite specific - a prepressurised hydraulic ram. For the energy cost in changing the spring coefficient and in operation (losses), you gain controlability, a failsafe state, and an easy way to turn the springs off.