Load-matching water pumping turbines

[Brian O Callaghan]

Hello Hugh and everybody,
My name is Brian O Callaghan and I am taking the MSc course at the CAT in Wales this year. I was asking Hugh about potential dissertation ideas, and after a few months of talking with the lecturers here, I have finally set the topic (phew!)

The title of the dissertation is 'Can dynamic load-matching improve the efficiency of wind-powered reverse osmosis filtration?'

I have attached two documents that hopefully will explain the topic a little. The idea is to change either the number of pumps operating or the displacement of a single pump according to wind speed in order to use the maximum amount of power from the wind at any given time. If such a system could work, the output would be multiplied several times, depending on the wind patterns. At the moment, I'm not sure if it is possible to do this without increasing the cost and unreliability of the wind pumps, although I do have a few ideas...

If any of you have any thoughts or know of any projects trying to do similar things it would be fantastic to hear them.

I am hoping to collect data in a few ways -
    - testing existing water pumps for efficiency and response to changes in wind speed
    - building a working model to test predictions of output

For the testing of the water pumps, I would need to find people willing to have measurements taken on their water pumps, ideally in Wales or nearby. I have asked several organisations, but haven't had much of a response, so if anyone had any tips on this it would be great.

I am living in mid-Wales myself for the next few months, and am always happy to help out with sustainability projects, so do let me know if anyone is in the area.

All the best,
Brian

[jorge....@gmail.com]

Nice Dissertation!  I was thinking of doing something similar many years back but never got around to doing something practical.  My plan was to have a turbine run a pump that would make the RO system work in variable speed.  Not sure if that is feasable with the RO setup, am not really an expert in this realm.  

I am doing wind-electric waterpumping tests based on a thesis done by Edward Charles Lemmer from Stellenboch University.  He designed a Hugh Piggott Stator to run on a 3HP pump.  We are in the process of bench testing the design and simultaneously have a turbine running.  Unfortunately our site has low winds, therefore we are looking for a 5m/s or better site for the field tests.  We are based in South India and RO systems are very common over here, so your work can be quite applicable here.

Hugh mentioned in his talk at the windempowerement conference that making a modified stator would be of interest.  Currently the stators are made with fiberglass casting process.  The Proven turbines have a pancake generator with laminated stators.  In Vietnam I noticed that the stators for picohydro turbines were not laminated but made of cast iron instead.  This was a cheaper alternative to laminations.  I am wondering if by developing a specially designed cast iron stator we could repower the generator and yet maintain some low eddy currents (to a manageable level!).  It would be very easy and cheap to make some cast iron stators here, but we would need some electrical engineer to make a nice design for testing.  I can imagine this "repowered" stator could be a great combination for pumping applications.

If you have trouble finding the thesis for the lemmer study I can email it to you.

Jorge Ayarza

Minvayu

Auroville, India

minvayu (at) gmail (point) com

[Brian O Callaghan]

Hello,
I am putting this question below the other comments on this project for context.
The dissertation I describe here is proceeding well, we have come up with a working system. The problem is, it is very complex, and therefore expensive and probably fairly unreliable. I am wondering if anyone could point me to open source code for Hill climb search MPPT? The plan is to use feedback from a flow meter to adjust stroke on a wind pump through a searching routine in an arduino module, much like voltage is used in wind and solar MPPT systems.
Any pointers on this would be much appreciated,
Brian

[Hugh Piggott]

hi Brian,

I wonder if you have considered using a lookup table for the pump?  It is generally considered a better solution for wind turbines.  The wind changes all the time and so it is not possible to invent an algorithm that learns the best loading as you can with PV or hydro MPPT.  Generally with wind the solution is to use a table of power vs voltage, and to construct this table for the machine using performance testing with an anemometer.

cheers

Hugh

[Brian O Callaghan]

Hi Hugh,

Just to clarify, this is for a mechanical wind pump. I see what you mean about fluctuations in wind, I was thinking that by averaging out the flow rate over the preceding minute, you could at least have a reasonable estimate of where the variable stroke mechanism needed to be. I should specify that this is for off-grid systems, so we're trying to keep motor consumption to a minimum. At the moment, we move the motor for 15 seconds every three minutes. So the idea would be that the arduiono would average the flow rate over 30 seconds or so, move the variable stroke mechanism up a bit for maybe 30 seconds (or enough to avoid responding to gusts and lulls), see what happens, if flow increases, it remains there, otherwise, it moves to a corresponding position downward. And repeats this every... five or ten minutes.

So, given that a rough approximation of the correct position of the variable stroke mechanism would be well worth it, do you think it would be worth pursuing?

We already have power curves for the turbine, but could you explain how you could use a look up table? By associating a flow rate with a position of the variable stroke (actuator) mechanism?

Thanks a million,

Brian

--
You received this message because you are subscribed to the Google Groups "General Discussion Forum (Wind Empowerment)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to wind-empowerme...@googlegroups.com.
To post to this group, send email to wind-emp...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/wind-empowerment/2a6f12f3-52aa-44b8-8b0a-892b05ddf431%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Hugh Piggott]

It seems to me that the logic remains persuasive that you cannot find the best stroke by trying to track it, if the wind is changing meantime.  I would suggest a table of stroke versus rpm or flow whichever is easier to measure.  I suspect that rpm will be the easier.  You coudl restrict the adjustment to once per minute or only if the error exceeds a certain % or a combination of delays and error levels to economise on control power usage.

On Wednesday, December 16, 2015 at 10:40:46 AM UTC, Brian O Callaghan wrote:

Hi Hugh,

Just to clarify, this is for a mechanical wind pump. I see what you mean about fluctuations in wind, I was thinking that by averaging out the flow rate over the preceding minute, you could at least have a reasonable estimate of where the variable stroke mechanism needed to be. I should specify that this is for off-grid systems, so we're trying to keep motor consumption to a minimum. At the moment, we move the motor for 15 seconds every three minutes. So the idea would be that the arduiono would average the flow rate over 30 seconds or so, move the variable stroke mechanism up a bit for maybe 30 seconds (or enough to avoid responding to gusts and lulls), see what happens, if flow increases, it remains there, otherwise, it moves to a corresponding position downward. And repeats this every... five or ten minutes.

So, given that a rough approximation of the correct position of the variable stroke mechanism would be well worth it, do you think it would be worth pursuing?

We already have power curves for the turbine, but could you explain how you could use a look up table? By associating a flow rate with a position of the variable stroke (actuator) mechanism?

Thanks a million,

Brian

On 16 December 2015 at 15:51, Hugh Piggott <scora...@gmail.com> wrote:

hi Brian,

I wonder if you have considered using a lookup table for the pump?  It is generally considered a better solution for wind turbines.  The wind changes all the time and so it is not possible to invent an algorithm that learns the best loading as you can with PV or hydro MPPT.  Generally with wind the solution is to use a table of power vs voltage, and to construct this table for the machine using performance testing with an anemometer.

cheers

Hugh

--
You received this message because you are subscribed to the Google Groups "General Discussion Forum (Wind Empowerment)" group.

To unsubscribe from this group and stop receiving emails from it, send an email to wind-empowerment+unsubscribe@googlegroups.com.
To post to this group, send email to wind-empowerment@googlegroups.com.