Solar Concentrator Reviews

Review of cost predictions by Gang Xiao and Response by Marcin

Hi,

This computation is not quite correct. I am doing a better calculation; I will let you know when it is OK.

Here I would like to give you my opinion on the cost analysis. I don't expect you to believe me right now; I only hope that you remember my estimations when you've learned it the hard way via a prototype whose efficiency falls short of your expectations.

Your cost computation on the linear fresnel concentrator is fundamentally flawed. It is a general understanding within the solar concentrator industry that the main cost of a concentrator is not the raw materials but the manufacturing cost meeting the high optical precision requirement. If you put a value of zero for the manufacturing cost in your cost analysis, it does not make much sense.

To my knowledge, my trough design is the only method that can avoid this fatal correlation between cost and precision. That's the reason why my cost is low. If your design is based on known technology, you can only reduce cost by sacrifying precision and hence efficiency, and you will end by getting a product with worse performance/cost ratio.

In fact, the cost to pay by using flat slat mirrors is exactly that the precision requirement is several times higher for a given concentration ratio. I've told you that you need a precision of plus minus 2 mrad, or 0.1 degree. At this level, everything is non-negligeable: manufacturing (fixation) error, deformations by gravity, wind, aging, etc. Your design should take all this into account, and calculate carefully incorporationg error margins. Of course, the cost will inevitably grow.

For example, using a simple screw to fix the mirror on the rod is much too coarse. The screws will get loose over time, and it is impossible to refasten on the field towards a precision of 2 mrad.

Such a precision requirement needs precision manufacturing (fixation) equipments and trained personnel, so you cannot put labor cost to 0 by counting on unexperienced volonteers. And I have made a preliminary cost analysis for you, according to the industrial standard. My estimation is about 200$/m^2 plus installation cost, in line with commercial products. What I am not sure is whether you will be able to get the same quality standard as a commercial product.

On the other hand, the steam engine net electricity output will be single digit if you use direct steam. My next precise calculation will show that it is around 8%. So the overall efficiency is less than 4%, so the net electricity cost is more than that of optimally tilted photovoltaic panels, if you add shade loss, end loss, cosine effect, as well as the miss of diffuse light. Your claim of 1$/W is much too exaggerated.

In fact, electricity-only solar thermal technology is only viable for large scale power plants, with high quality concentrators and sophisticated turbines and so on. But this market is for the time being neither for you nor for me: it is not a technical problem but a social one. It is big money affair and political lobbying, we have no access to it.

We can only look at small scale applications. But even with my trough design that is much cheaper and of much higher performance, I don't dare advocate for small scale electricity-only applications. This is not economically competitive. At small scale, CHP is the only competitive solution.

It is here that the compactness of my design shows its great advantage. CHP means that it must be installed near heat users, because long hot water pipes mean huge heat losses. Ideally, installations for individual homes. For this, linear fresnel is downright impossible, because of its minimal practical size. You need a minimal field size of tens of square meters to minimize endlosses. How can a family do to consume the huge quantity of collected heat? In most places, the only real "market" is winter home heating. However, linear fresnel collectors simply don't work during winter due to a too low sun angle. The cosine effect is too high for flat north-south installations, and shade loss is too important for flat east-west installation. The incoming energy will even not make up for heat losses. I call this phenomenon "solar hibernation". The only solution is to make tilted installations, but that's explosion in cost. In most residential areas, the height of a tilted linear fresnel collector would simply exceed what is allowed by the legislation. And you will have hard time fighting wind load and so on.

Response to Critique

Yes, but how many industrial players are working with flat arrays? That coudl be the secret to low cost - via DIY - as the intelligence is taken out of the manufacturing process and put into the advanced electronic controls. Isn't it true that field adjustment of alignment, slat by slat, will be a foolproof way to obain the required alignment?

The only trick is - a properly stiff mounting structure for the slats. That does not appear to be an intractable problem, and these guys from the UK claim to have done it at low cost -

http://openfarmtech.org/weblog/?p=446 We are allowing for 50 cent per watt labor costs.

Are you saying that oil heat exchange is the answer?

The electronically valved steam engine is reportedly superior to turbines in the sub 500 hp power scale.

What do you think of the potential of a system like Nolaris? (http://nolaris.ch/17-0-Our-Technology.html)

Marcin

Further Communication

   Why is it not possible to have a secondary reflector at the  collector tube, with nonimaging optics, for another factor of 3 of concentration onto a 2" tube, whe we start with 6" slats?
   Total theoretical limit becomes 48x concentration.

This was my unit confusion. In theory, what you do is possible (and it's the limit), but this only gives you 16x concentration not 48x. The geometric concentration ratio means mirror surface devided by by the receiving surface, so you should calculate with the circumference of the tube.

In practice, you have to incorporate error margins and increase the tube diameter (or circumference) by 20% or so, because reflections will never be exact. That's why I give you a practical concentration ratio of 13 times.

However, with more careful calculations taking into account the sun's angular diameter, I am afraid that 20% is not enough. It should rater be 30%-35% or so. That's very bad. If you give me all your design diameters (gaps between the slats, distance between the slats and the receiver), I can tell you how this computation should be done.

---

   We are aiming for a working temperature of about 350C, and we will have to do as many slats as are necessary to achieve that. Then, the question is, how much power will be available to the steam engine, if we optimize for a temperature of 350C.

You need a much higher concentration ratio to reach that temperature. At least 20 times geometric. This would mean roughly at least 30 slats and a surface error of 1 mrad. If you don't believe me that this is impossible, just try it.

And you have to insulate the receiver with vacuum technology. Glass wools and double panes won't be enough to check conductive and convective thermal losses.

On the other hand, 350C with big turbine can give you up to 30% net heat-to-electricity efficiency. This is the same for the big trough power plants. Do you know any existing linear fresnel powe plant? If yes, please read their cost report. If no, there is certainly a reason.

---

Marcin Jakubowski wrote:

   Yes, but how many industrial players are working with flat arrays? That coudl  be the secret to low cost - via DIY - as the intelligence is taken out of the manufacturing process and put into the advanced electronic controls. Isn't it true that field adjustment of alignment, slat by slat, will be a foolproof way to obain the required alignment?

My own experience tells me that this kind of argument won't convince anybody. Even showing the prototype won't be enough. You have to be able to sell your product with the promised price.

Now there is no fundamental IP in the linear fresnel design. What makes the industrial prices is mainly the real manufacturing cost.

Field adjustment means heavy equipment. The big troughs align to 3 mrad precision, and heavy and special equipments are used for that. Did you notice Sandia making big hypes on an invention that improves this process? This is definitely not suitable for DIY. Don't forget that your adjustment should last 20 years.

One of the reason why people use big troughs is that the alignment (adjustment) costs too much. So big troughs mean less adjustment, hence economy.

   The only trick is - a properly stiff mounting structure for the slats. That does not appear to be an intractable problem, and these guys from the UK claim to have done it at low cost -

   http://openfarmtech.org/weblog/?p=446

I've seen that. No data to show what is their performance and quality, so that doesn't mean much.

      If you put a value of zero for the manufacturing cost in your cost
      analysis, it does not make much sense.

   We are allowing for 50 cent per watt labor costs.

What it corresponds for one slat? What is key to me is how you plan to do to meet the 2 mrad error rate. And what's your calculation with wind deformation and so on. This is not field adjustable without heavy equipment and a proper design!

      To my knowledge, my trough design is the only method that can
      avoid this fatal correlation between cost and precision.

   Tell me more about your key to success.

The key is to obtain the desired shape entirely thru natural elastic deformation of flat material. The problem is, whenever you try to force the material to follow your predefined shape, you get a cost to precision correlation. But natural deformation can give the high precision with low cost, if you know how to induce the deformations. There are quite some mathematical computations behind the seemingly simple construction.

      On the other hand, the steam engine net electricity output will be
      single digit if you use direct steam.

   Are you saying that oil heat exchange is the answer?

No, the key is temperature hence concentration ratio. Of course, with high temperature, you will have to switch to oil, for otherwise the pressure would be too high.

      In fact, electricity-only solar thermal technology is only viable
      for large scale power plants, with high quality concentrators and
      sophisticated turbines

   The electronically valved steam engine is reportedly superior to turbines in the sub 500 hp power scale.

Don't buy into people's commercial hypes! Always ask the question of performance to price ratio.

I am doing some detailed steam engine calculations. It seems that it's better than I first thought; but I haven't finished yet.

   Shade loss fraction may increase, but can be potentially mitigated by sufficient inter-slat spacing.

That's too quick. Increasing inter-slat spacing proportionally reduce concentration ratio or reduce the error tolerance, either one or the other, or both.

   This may work in winter - and it can definitely work under the assumption of an optimally-matched steam engine. At the very worst, there may be a limited range of performance in winter. At the very least, winter operation could afford space heating.

   What do you think of the potential of a system like Nolaris? (http://nolaris.ch/17-0-Our-Technology.html)

No idea because there is no technical information nor cost. There are too many things like that on the internet, most of them are vaporware.

You know that a person with marketing talent may successfully sell a technically inferior product. This kind of website is part of such talent.

I once contacted a trough manufacturer who told me that his product costs only less than 200 euros per m^2. I never know his exact price, but from his ads, I can guess that it's probably over 1000. Be careful.

I myself am suffering from this situation. Nobody believed my cost analysis. So now I try to make the product and sell it for the promised price, with promised quality. I suspect that even this is not enough. I will have to make the product test by third party and so on.

Your project will be in the same situation. Marketing hypes can bring some unexperienced volunteers, but knowledgable people will only be convinced if you show them concrete product with complete third party test data. In between, if you don't have the capability to fix every technical detail, you will fail.

With respect to your project, I am in a different situation. I can guess what are the problems you will encounter. For the time being, you haven't convinced me that you will be able to solve them. So I am not very optimistic about your linear fresnel. Of course, if you finally succeed, I will only be happy with that. I am not really considering you as a competitor: our technologies are somewhat complementary.

Getting Interesting

On Tue, Feb 24, 2009 at 3:11 PM, XIAO Gang <xiao@unice.fr> wrote:

   Marcin Jakubowski wrote:

       Have you ever tried to align a single reflector onto a receiver tube? I don't see why that is overly difficult, if one has closed loop feedback.

   Can you image how many calculations I have done, and how many tests I have made? 

Show me the data. There would be no need for me to do the experiments if you provide some data.

   I know what is realistic and what isn't.

       Now, our strategy is to sacrifice performace at the gain of absolute lowest cost. We believe that this point is a winner for overall cost competitiveness.

   Except that accumulation of performance loss will end at a worse performance/cost ratio.

That may or may not hold true, depending on the details of the situation. Yes, typically lower performance means higher cost/performance, but if cost is reduced more than performance is reduced, then we may have a winner. That is not impossible, as you suggest.

   The whole picture is linked. I am certain that many people have had the same idea. The linear fresnel design existed at least since 1970s. 

We didn't have cheap tracking electronics until the 10 years or so back, though.

   The commercial products are usually optimized for the performance/cost ratio. If you can break this optimization without non-trivial inventions, all these engineers are good for nothing.

Correct.

   Well I don't say that's impossible, but the chances are rare. In most of the time, you will be reiterating errors already encountered by others.

Exactly. That's why before any experiments, I am and will seek the advice of those who have more experience. That's the nature of the open source effort.

   XIAO Gang (~{P$8U~})                      http://wims.unice.fr/xiao/

More Issues

              Have you ever tried to align a single reflector onto a
              receiver tube? I don't see why that is overly difficult, if
              one has closed loop feedback.

          Can you image how many calculations I have done, and how many
          tests I have made?

       Show me the data. There would be no need for me to do the experiments if you provide some data.

   I am giving your the compiled data, but you are sceptical right now. For example, that 2 mrad precision is not a DIY affair. But here the test is easy to mount.

If we have a secondary nonimaging optics reflector shroud (like in Powerfromthesun.net) in front of the collector tube for 2x secondary reflection, then at 10 meter separation from reflectors, we can accept 6 mrad presision.

With individual slat motors and 60 pitch gear of 2" diameter for controlling slat rotation gets you 2 mrad accuracy.

What is so difficult about implementing this in practice?

I'm interested in hearing more about your experimental procedure with which you obtained your precision results so I can assess your statements properly.

          The whole picture is linked. I am certain that many people have
          had the same idea. The linear fresnel design existed at least
          since 1970s.

       We didn't have cheap tracking electronics until the 10 years or so back, though.

   It's still not yet zero today. So what is your cost? Where is the schematics, parts list, cost addup, programming and so on? By the way, do you know that you should use stepper motors? Brush DC motors don't last long.

Don't have any of it. All I know is that we can use cheap off-shelf open source controllers like Arduino, tracking algorithms can be implemented, and if need be, we'll tool up for fabricating our own stepper motors if that turns out to be prohibitive. I don't see any a priori reasons that make these challenges intractable. You must remember that we will go as deep into developing tooling as needed to achieve required cost predictions.

          Well I don't say that's impossible, but the chances are rare. In
          most of the time, you will be reiterating errors already
          encountered by others.

       Exactly. That's why before any experiments, I am and will seek the advice of those who have more experience. That's the nature of the open source effort.

   Sorry but when I look more closely at your design, I've just got the impression that it's not yet a question of performance/cost: it simply won't work. Too naive.

   Secondary reflector. How would you do to protect the metal mirror surface? Plastics would melt down, glass would crack under this temperature. Left unprotected, the metal surface would oxide in just a few days, under the accelerated oxidation of high temperature. Other people put the secondary mirror under evacuated environment, but that's again not a DIY affair, and the cost is probably prohibitive if you can find somebody making it.

I don't know yet. If we can't solve this issue, we will have to go to additional reflector slats to make up for this.

   Glass cover of the receiver. During our trough test, people once cracked a glass cover with a temperature gradient of only about 70C. If you want to reach 200C for your receiver, you have to put margins and test it under 250C or even 300C. At this temperature gradient, even a borosilicate glass of the highest grade won't resist. You'll need the kind of special glass found in kitchen ustensils and ovens, but you have to incorporate the proper price tag into your cost estimate (it's going to be expensive). 

What is the going price estimate for this glass? We may have to produce this glass ourselves it it is not cheap off-shelf.

   Or you should switch to evacuated tubes, 

Evacuated tubes - I don't think that will have the same cost performance as non-evacuated conditions.

   or reduce your temperature objective to less than 150C. Of course, 150C won't give you much electricity. If you want 300C or so, evacuated receiver is a must.

   It's going to be a subtle problem. When there is end loss, the receiver is half hot half cold. If you don't take care, even a heat resistant glass cover could crack in the middle. That's one of the reasons behind receiver tube breaks in the big trough power plants. Of course, people know the solutions, but these are trade secrets.

   And there are norms for solar collectors: wind resistance, hail resistance, lifetime, etc. If your product doesn't respect these norms, it's no better than a toy. So for example you have to compute/test the mechanical behavior of your slats/receiver under the wind pressure of 120km/h, as well as the deformation extent under 60km/h, etc. The low profile is not an excuse.

Agreed.

Marcin

Mirror Testing

Hi,

By curiosity, I've just made a small mirror test for you.

I used a high quality vanity mirror which I picked for earlier optical experiments. Excellent surface flatness.

The mirror is of width 13cm, with an almost normal angle to the sun. At a distance of 5m, the image scattering reaches 21cm. That's just the image width, not counting positionning errors. Therefore for this distance and with slats of 15cm wide, the minimal receiver width should be 23cm, or the reflected light will be partly lost. Adding inevitable error margins, 25cm or 26cm is a strict minimum. That's even worse than what I calculated. With 16 slats, the concentration ratio won't exceed a single digit.

Adding slats is no big improvement, because the distance should increase, and the image wider. So a wider receiver has to be used, and the concentration ratio almost remains the same.

An even bigger trouble is with deformation. My mirror is only of 20cm long. At this length, a torsional force of only a few dozens of grams is enough to deform the image by several centimeters. Now even a mild wind force is of 10kg/m^2, and at equal tortional force, the deformation is proportional to the square of the length!

Try it for yourself.