Monday 24 June 2013

Peter Fraenkel on fundamental technology challenges - - - - - - - - - - - - - - - - - - a discussion




The fundamental challenge of large torque

At the quickfire seminar series at All Energy 2013, tidal energy veteran Peter Fraenkel (Fraenkel-Wright) gave his views on fundamental engineering challenges facing both tidal stream and wave energy: the lower the speed of the prime mover, the higher the torque, and the bigger the machinery. He estimated that tidal turbines were an order of magnitude slower, and WECs (wave energy converters) two orders of magnitude slower, than fossil fuel turbines.

Peter Fraenkel’s summation of the fundamental engineering constraints carried the implicit suggestion that wave energy would always be more expensive than tidal energy. Perhaps some insight might be gained from examining the reasons why tidal energy has progressed faster than wave energy. Of course I naturally feel compelled to defend wave energy as a matter of principle, so while I acknowledge that tribalism is a motivating factor for me, I shall do my best to prevent this from affecting my objectivity.

There are two issues surrounding low speeds in wave energy: one is the provision of PTO (power take off) equipment that can deal with large torques, and the other is the requirement for large primary wave activated bodies.

Do high torques necessarily make wave energy expensive?

When the surfaces acted upon by the resource move slowly, the first stage PTO system must be strong enough to deal with high torques. In general these are large and expensive, and this would seem to indicate that wave energy is facing higher hurdles than tidal stream energy. Of course, another difference between these technologies is the directionality of forces in the first stage of power capture: for wave energy the forces on the PTO system are oscillatory and generally bi-directional, whereas for tidal stream the forces are unidirectional. Wave energy has the additional challenges of rectification and frequent operation at part load. Most PTO equipment (particularly electrical generators) are very inefficient when operating at part load.

If the first stage of PTO in a WEC was an electrical generator, then the problems of high torque, rectification, and inefficient operation at part load would result in expensive and unreliable (because gearing is a weak point) power generation. There are however alternatives to electrical PTO which would be better suited to wave power. There is an existing supply chain for hydraulic drives that are suited to high torque, low speed applications. Rectification using hydraulic drives is a well proven method, and hydraulic drives are widely considered to be more robust than systems comprising gears, electrical generators and power electronics.

Hydraulic systems are not immune to the problem of lower efficiency operation at part load. Technology initially designed to mitigate the part load problem in wave energy has been commercialised and successfully implemented for other engineering applications. Artemis are now developing a high part-load efficiency hydraulic drive for use in their parent company Mitsubishi’s large offshore wind turbines. Large scale wind turbines are another technology facing problems due to high torques. As torque increases with size, this is seen as a potential barrier for scaling up to even larger turbines. If the development costs of marinising this technology are funded by offshore wind, this puts wave power (and tidal stream for that matter) in a good position to benefit from this technology.

It is also interesting to note that for a WEC, it is not compulsory for the primary wave activated body to be directly damped in the first stage of power capture. It is possible for some form of gearing to be implemented prior to power capture. A well known example of gearing that is built into a design is an OWC (oscillating water column). The primary wave activated body is the column of water oscillating within a caisson. The surface of the water column activates the column of air above it. Before the air passes through the turbine, the internal shape of the caisson narrows, and as a result the air speeds up. Thus a system can be designed so that the PTO operates at a higher speed than the primary wave activated body.


Do big bits necessarily make wave energy expensive?

The other problem with slow motions is the requirement for big surfaces for the resource to act on. In tidal energy, the moving parts acted upon by the resource (blades) are fairly sophisticated and expensive bits of engineering. In wave energy, the moving parts acted upon by the resource are relatively unsophisticated. Indeed, they are typically much larger than tidal turbine blades, but the difference in cost is not as large as the difference in mass. The primary wave activated body for a wave energy converter (WEC) is typically a large rigid body, and much of its mass could be gained from cheap ballast. The shell of the body has fewer requirements for precision engineering than a turbine blade.

There is certainly an opportunity for cost savings in wave energy by considering other materials for the primary wave activated bodies. It is interesting to note that Pelamis have announced that they are moving from steel to concrete tubes for cost reasons (Andrew Scott, speaking at the Wave&Tidal 7 session at All Energy 2013).

A key issue for WECs is that the size of a device is related to the size of the waves that can be caught. The larger the size of the wave activated body, the broader the bandwidth of capture, the longer the wave period at which capture efficiency begins to drop off, the higher the loads induced, and the higher the wave height at which load or power shedding takes place.

Clearly the size of a WEC with respect to a particular resource is an important determinant of cost of energy. Unfortunately not enough is known yet about the scaling laws for the cost effectiveness of WECs. Those who have considered device size (relative to resource) in cost models acknowledge the uncertainty due to insufficient validation data, and there is still plenty of debate about the most economically viable size. The group who have access to the most validation data for their cost model are Pelamis. Although results from their cost optimisation program have not been published, at All Energy they did give an indication of where the next generation Pelamis (P2e) was going. In short, the developer with the most operational and iterative design data thinks that, for the Scottish resource, they can bring down the cost of energy for their device by using larger wave activated bodies. The only thing we can say with certainty about economic scaling laws for wave energy is that they differ to those for tidal energy.

It is interesting to break down what advantages are bought with large wave activated bodies: broader bandwidth, higher resonant period, higher induced loads, and higher load shedding limits. This presents an opportunity for wave power: are there other, cheaper, ways that we can attain these goals? There is evidence that bandwidth can be broadened by using a multibody system, or capturing power from a DoF that is made up of two modes of motion (e.g. the sloped IPS buoy compared to the heaving IPS buoy). Power capture from more than one DoF also appears promicing. It has already been shown that power capture in heave has less potential than heave combined with surge or pitch. I would be interested to hear about academic work that explored the practical (using a non-reactive, damping only PTO) power capture potential of a wide range of combinations of DoF for single and multiple bodies. The question of whether there are ways to increase the allowable motions before load shedding comes into play follows on to another fundamental challenge raised by Peter Fraenkel.

The fundamental challenge of the average to extreme ratio

At the recent RenewableUK conference, I asked Peter Fraenkel why he thought wave energy was lagging behind tidal stream in terms of cost of energy. He answered that the extremes of a resource dictated the project costs, while the average of the resource determined the energy generated (income).

If this issue is indeed a fundamental cost driver, then the wave power industry should be giving this more consideration. Here are some questions that arise from this fundamental engineering challenge:
  • We need to be able to express the project costs as a function of resource extremes. Is the maximum wave power transport directly related to the maximum load? There is anecdotal evidence that for WECs the highest loads occur at steep waves at the resonant period, which might occur during power generation, rather than during storm conditions at longer wave period where the device is less resonant. The questions that arise from this is whether peak loads can be reduced by hydrodynamic load shedding, or operational mitigations?
  • We need to be able to express the revenue generated as a function of resource average. This is strongly dependant on the size of the device, but for a fixed size, nameplate rating is important. As the rating has an impact on capital costs, the choice of rating, and the resulting capacity factor, amounts to a CoE consideration. We do not yet know what capacity factor will result in optimal CoE - it might depend on the resource and the type of device. For a fair comparison of CoE in medium and high power resources, capacity factors optimised for the resource should be used.
  • Are there ways to operate a WEC that could increase average power generation relative to maximum loads? For example, would it be worthwhile in low power sea states to actively moor a WEC in a high performance configuration that attracted high loads, while in high power sea states allow it to passively weathervane into a low performance configuration that shed loads? Perhaps there is an opportunity for semi-flexible materials that transmit low and medium loads, but deflect rather than transmitting high loads?
  • It has been observed that sites with medium wave resource have a smaller gap between average and extreme resource. Do CoE models presently represent this difference? It would be useful to develop an understanding of how much the CoE is impacted by the ratio of average to extreme resource. A higher ratio could result in higher design tolerances, high factors of safety, reduced weather windows, and increased maintenance costs.
  • Anecdotal evidence suggests that semi-sheltered testing sites have a greater ratio of average to extreme resource. If this is indeed the case, then this could result in overly pessimistic CoE estimations.
  • If it is found that the average to extreme ratio has a significant bearing on CoE, perhaps the industry should consider developing early arrays at medium power sites. There is an additional synergy with early commercial projects for medium power sites: as large WECs are resonant, and medium power sites probably have lower period waves than high power sites, the most economical size of device in a medium site will be smaller than that for a higher power site. To incrementally reduce risk and increase technology readiness, it is more cost effective to use smaller devices for early projects and gradually increase device size with subsequent projects. This trend in upward sizing has been seen for both Aquamarine (Oyster 1->Oyster 800) and Pelamis (P2->P2e). This suggests that the early prototypes of both companies may have produced better CoE if designed and operated in medium power sites. Although power production would have been lower, there would probably have been cost savings on lower rated PTO equipment, as well as installation and maintenance.

Final comments from Peter Fraenkel

Peter Fraenkel responded to the ideas discussed above with a few additional thoughts. He encouraged an openness to new ideas, and a focus on overcoming the challenges of high torques and a high ratio between the average and extremes of the resource:

"there is always the chance someone comes up with a wave energy conversion system which is such a game changer as to overcome the fundamental disadvantages."

He thought there was an opportunity for devices that were simple, had low capital costs, or made use of alternative materials. Yet he also thought that as the technology matures, design complexity will be needed to make wave energy cost effective. He offered a comparison to another technology:

"Engineering developments have always been cleverer than anyone expected - even the designers of the first modern airliner - the DC3 - would have found a Boeing 747 or an Airbus A380 difficult to comprehend even though they share similar construction and principles of design with the DC3. It only took 30 years to go from the DC3 to the 747 yet DC3s are still in widespread use nearly 80 years after they were designed. But that is all a slight distraction from wave power other than to make the point that we will probably eventually be surprised at what becomes possible."

He gave an example of the type of innovation needed to overcome the fundamental technology changes facing wave energy capture. He thought that gearing implemented before power capture was important, as it would allow the use of higher speed, hence smaller and cheaper, PTO equipment. He suggested a primary absorber of reinforced rubber, which transmitted forces across a large area to compressed air, which would be damped when passing through a small area. Such a system would allow gearing, is fatigue resilient, and would not be as heavily stressed as rigid steel. He observed that several developers are already working on such systems.



Related posts:

http://www.wavepowerconundrums.com/2013/02/nutshell-technology-readiness-performance-matrix.html
http://www.wavepowerconundrums.com/2013/04/opportunities-for-sites-with-low-wave-resource.html
http://www.wavepowerconundrums.com/2013/04/iet-towards-commercialisation-seminar.html


Image credit:
'Sublime' by badjonni: http://www.flickr.com/photos/badjonni/3844800850/in/set-72157594290716209




13 comments:

  1. The conversion of waves into power does need big devices but like you I am optimistic that they can become cost effective. I am however concerned that devices which work well at model scale may not scale up well, and conversely successful full size devices might not look good at model scale. Our school took part in the Junior Saltire competition and won it in 2012 with a Cockerell raft type device driving a large diameter alternator. In 2013 we got second with a heave raft/linear alternator. This, though large for the output, had only one moving part. The device made a surprisingly large voltage at low linear speeds.

    ReplyDelete
  2. hi Topher. I'm very excited to hear that schools are doing wave power design projects - good on them! Scaling can be a problem if the WEC uses a gas as part of the system: the pressure / volume relationship does not follow Froude scaling (very inconvenient!): http://en.wikipedia.org/wiki/Froude_number It could also be a problem if the experimental model had lots of losses, e.g. friction between solid components - bearings or sliders- or between the device body and the sea - viscous losses caused by sharp edges on a model. With larger devices there will be a bigger budget, so one would hope that lower friction components would be used! Viscous damping is less of a problem with large devices. Apart from the issues of viscous losses, mechanical friction and representing gasses, modelling WECs at small scale should be possible if Froude scaling is used. Of course there are other reasons besides scaling why tests of small scale models might not be representative of the real thing: these are to do with what simplifications are made in the tests: typically small-scale models would be tested in waves with one direction, and with one type of spectrum - such as 'Jonswap', or even worse: regular waves (approaching sinusoidal). The way that experiments are conducted could have a big influence on the results.

    ReplyDelete
    Replies
    1. Is there not the same conflict as with testing ship models, i.e. it is seldom possible to keep Reynolds No and Froude No the same for model and full scale? Some of our devices use baffle plates as reactors and they will be Re sensitive. There are probably dimensionless numbers related to the electrical generator too.

      Delete
    2. Yes, the same conflict arises: you must choose whether to keep either the Reynolds No or the Froude No fixed. If the WEC was powered by lift (hydrofoils) you would choose Reynolds scaling. If developers wanted to check the impact of a different Froude number, they might use model tests to validate a CFD model of the scaled test model, and then see what the CFD model gave for the full scale dimensions. If the WEC was powered by drag (most are - e.g. oscillating rigid or flexible solid bodies, or oscillating water columns) then you would choose Froude scaling. If developers wanted to check the impact of a different Reynolds number, they might experimentally measure forces that do not Froude scale, then run a numerical simulation of the motion at both model scale (using the measured forces) and at full scale (using estimations of the full scale forces).

      With Froude scaling, the viscous losses will be proportionally greater at model scale. Any sticky-out bits (like your baffle plate) will be loosing way more energy, and as a result will be a lot more slowed down, than at full scale.

      The question is, will this result in an underestimation or overestimation of power capture? It depends on the phase difference between the motions of the two bodies between which there is power take off. If the motions are out of phase, scale models will underestimate power capture, e.g. imagine an attenuator like Pelamis with a star-shaped cross section - clearly a bad idea for many reasons; the point being that model scale tests will underestimate power capture. If the motions could be in phase, such as relative heave motion, then scale models will underestimate power capture. In the scale model the baffle plate might provide a nice stable point of reaction, whereas at full scale, it would be more likely to track the motion of the body it is supposed to be providing reaction for.

      As for electrical generators, frictional losses would be proportionally higher in scaled model tests. This would lead to an underestimation of power capture.

      Delete
  3. Hi Alexandra,

    Only just discovered your reply, sorry to be so unobservant. The Cockerell raft was first tested in real sea waves within welly depth of the shore, so they were complex and irregular. It still worked well. But the test for the competition was in a long narrow tank at Strathclyde Uni with the simplest possible waves, sinusoidal, single frequency and linear.

    The heave raft/linear alternator had too much mild steel in it to risk sea water, so it only went in the tank. I think its resonant frequency could have been altered electrically by changing the way the output was handled.

    Both these devices used a lot of expensive materials (magnets, copper) for the output but did have some promise.

    The current schools competition is for a tidal stream generator and I have two teams of 12 year olds making a horizontal axis device based on a car radiator fan and a vertical axis device driving the bicycle wheel alternator which was on the Cockerell raft.

    I have to admit I'm still more interested in the wave power problem than tidal stream, and am wondering about a new concept. You know the simple toy made from a big button and a loop of string? the string is stretched between the hands and the button is made to whizz round at high speed by pulling the hands apart. Then the button's inertia causes it to wind the string loop up the other way and then the hands tension the string, reversing and accelerating the button the other way.

    If one half of this toy was being stretched by wave surge, and the button was replaced by a generator and flywheel with a one-way ratchet drive, it would be a cheap and simple way to gear up the motion. The return torque to rewind the string could be pre-wound in the string itself. The generator would be mounted on a raft optimised to generate a combination of surge and heave forces, in the twisted tether which would connect it to its baffle plate reactor, and then its mooring.

    Two problems present themselves; the mooring would have to generate the average force of the wave, even if the baffle plate took the peak loads, and making sure the power output cable did not tangle with the tether might be difficult.

    Another toy which might be worth looking at is the pump action spinning top, which has a twisted rod which is pushed repeatedly by the operator to run the top up to speed.

    Anyway I may experiment with the twisted tether myself.

    T

    ReplyDelete
    Replies
    1. Your twisted tether idea is interesting. I'd not thought of it before so I will try and get my head around it. I had initially thought you may be talking about the mechanism used in a yoyo - there is a company who is looking into this: http://www.boltwavepower.com/
      Now I think you are talking about the following:
      https://blogs.glowscotland.org.uk/wl/KirknewtonPrimarySchool/2013/03/27/p1-old-toys-museum-handling-session/

      My understanding of this bit of string is that it is a combination of a gear (a small translation becomes many rotations ) and a spring (the kinetic energy transferred to the button/generator equals the potential energy in the coiled string). Here are my thoughts on such a mechanism:

      1. It could be that the energy stored in the spring is more than the kinetic energy put into the system by moving the hands apart (surge motion in a WEC) .
      2. In the button toy, it would eventually unwind itself (its equilibrium position), and then no amount of separating one's hands would wind it up again. In such a system, for a WEC, there is no direct path of power flow between surge motion and the generator's spin.
      3. You mentioned the system equilibrium would be in a pre-wound state: at the equilibrium the system stores no potential energy. A surge motion which increased the string tension would bring the system out of equilibrium - so this would work. Note the system would cycle around the equilibrium and it would reach the maximum winding stiffness in one direction before the other.
      4. Power can only be transmitted from translation in one direction (resulting in tension) - rope can't transmit loads in compression. I would consider such a system as half-rectification.
      5. I suspect that the repeated twisting of the cord in alternate directions might not be conducive to longevity!

      Although I don't think this will be an economic solution, it nevertheless makes for an interesting investigation, because it combines gearing (before power take off), transfer from translation to rotation, and short-term power smoothing in a simple mechanism, all of which are desirable for a cost effective WEC.

      Point 2 is the key thing here: how does power get from the wave activated motion to the generator? Your suggestion of the top would work. There is a rack and pinion mechanism turning a translation into a rotation (gearing is incorporated by choosing the size of the pinion). If I understood correctly, power is only transmitted from translation in one direction? If so, then this would be considered half rectified power transfer. An additional challenge is fully rectification: power transfer from both directions of translation!


      Delete
    2. Just read this in November! The 2015 competition is for a floating WEC so the idea is still of interest to us. The idea is to connect the top end of the twisted tether to the float mounted generator via a ratchet so the generator only goes one way. The tether has pre-twist put into the strands so its equilibrium position is twisted, like rope. During the power stroke the wave lifts the float, stretches the tether against the nearly still baffle plate, and it untwists, driving the generator through the ratchet. When the speed of twisting falls away at the end of the stroke below the generator speed, the ratchet disengages and the tether slackens and starts to retwist and contract in length. Then the cycle continues.

      Delete
  4. This comment has been removed by the author.

    ReplyDelete
  5. This comment has been removed by the author.

    ReplyDelete
  6. I'd be interested to hear (or see pictures) about how this project goes. Previously I'd pictured hands moving apart horizontally, so I thought power capture was in surge. Now I'm picturing the string being stretched vertically - so power capture is in heave - is that correct?

    I can think of three bits of design advice, which you may able to incorporate (or not). Firstly, it would be desirable to use really stretchy rope: it should be able to stretch roughly the same sort of length as the buoy moves when unmoored. This is because, for the same amount of power, small motions result in big forces. Big forces require more strength, structure and cost. Not that the cost of an extra thick rope is crucial in an educational model, but a snapped rope might be discouraging! Secondly, the baffleplate will probably not be stationary, as it too will feel the rope tension. If the baffleplate and buoy are vertically aligned, then a small baffleplate will tend to move almost in sync with the buoy above it, so the rope will not stretch much. So it is worth considering a drag plate (horizontal disk) that is shaped and sized so that only a very big force moves in the direction of the rope. Thirdly, if the baffle plate has small holes in it to baffle the flow, this will dissipate energy, and will not improve efficiency.

    Good luck with your entry!

    ReplyDelete
  7. Thanks for the reply! As a nonelastic twisted pair of strands untwists it gets longer, since the spiral path of the centreline of each strand has fewer turns. So there is a mechanism for the tether to expand without needing to stretch elastically, and elastic strands would absorb energy which would not necessarily be returned.
    Early tests indicate the need to space the strands further apart than they lie naturally, so either a thick sheath or a string of beads spring to mind, or a series of crossbars like the DNA molecule. Friction and abrasion issues.

    Baffle plate is as big as the rules allow, which is 550 mm square. There may be some parachute shape which could increase the drag coefficient. In my mind the device takes up a slightly inclined attitude with the tether about 20 degrees from vertical, but we will find out. Thus it's mostly a heave device but with some surge.

    ReplyDelete
  8. Also the gear ratio varies in a useful way. At the start of the power stroke the tether is fully wound and tension translates as a large torque. As the tether unwinds the helix angle gets larger and the gear ratio increases, which would be appropriate for an accelerating generator. Then the drive stops while the tether rewinds, and the generator slows down, ready to be accelerated again by a re-wound tether.

    ReplyDelete
  9. I found your blog on google and read a few of your other posts. I just added you to my Google News Reader. Keep up the great work Look forward to reading more from you in the future.

    Load Cell Manufacturers

    ReplyDelete

Comments

Note: only a member of this blog may post a comment.