Currently I'm building my own 5kW PV polar tracker and I need an advice on shading problem. To avoid shading it's necessary to implement special backtracking algorithm such that tracks solar panel backwards early morning and late afternoons.
A little explanation: http://www.lauritzen.biz/static/solutions/backtracking.pdf
Excample PV polar tracker itself:
Space your panels further apart, or on an axis where they won't overlap.
On the other hand, what exactly is the question?
If you have two panels like in the PDF you could measure the output of the two of them and adjust the angle until you get the maximum power. I suspect the concept could be extended to a large array.
Maybe you could add a small photodetector (light dependent resistor ?) to each panel to make it easy to detect shading without the trouble of measuring power.
...R
Robin2:
If you have two panels like in the PDF you could measure the output of the two of them and adjust the angle until you get the maximum power. I suspect the concept could be extended to a large array.Maybe you could add a small photodetector (light dependent resistor ?) to each panel to make it easy to detect shading without the trouble of measuring power.
...R
Ooh! Very nice idea! But I wonder how to measure single PV panel without any damage when its connected together serially. Because operating peak voltage is very high, e.g. ~700V.
nebulae:
But I wonder how to measure single PV panel without any damage when its connected together serially. Because operating peak voltage is very high, e.g. ~700V.
That's why I made two suggestions 
And here's a third ...
With a 5kW solar farm it would probably be good economics to have an extra pair of panels mechanically coupled to the rest but electrically separate from the rest and used only to detect and deal with the shadow problem. And another way to think about it is that if it is not worth buying 2 extra panels for this purpose then the shadow problem may not be not economically worth solving.
...R
Robin2:
nebulae:
But I wonder how to measure single PV panel without any damage when its connected together serially. Because operating peak voltage is very high, e.g. ~700V.That's why I made two suggestions
And here's a third ...
With a 5kW solar farm it would probably be good economics to have an extra pair of panels mechanically coupled to the rest but electrically separate from the rest and used only to detect and deal with the shadow problem. And another way to think about it is that if it is not worth buying 2 extra panels for this purpose then the shadow problem may not be not economically worth solving.
...R
Hehe
Every panel costs $500 $)
Actually your second idea is also good for this purpose. I can imitate panels with photoresistors.
nebulae:
Every panel costs $500 $)
And how much is the expected annual income? - cost on its own is not informative.
You don't need to check for shading very often. Perhaps you could take two panels off-line for a minute or 5 once per hour so they could be used to check shading?
...R
A few aspects:
You clearly do not need extra solar panels to do this, two or three LDRs mounted at top(, centre) and bottom of a representative (mid-array) panel would suffice, or perhaps two or three panels in case of something odd happening to individual panels (birds roosting?).
The shading problem is a combination of the angle of both panels involved, so all rows should track together; there is no point in for example, treating the front or back row differently.
All tracking algorithms are designed for slow response, so as not to be "tricked" by shadows or clouds.
You have the choice at the outset, of using "sun-following" sensors, or systems based on a calendar and clock. If you choose the latter, then it will also perform the "shading" algorithm without sensors (because it knows the effect of the sun angle). In fact, the same applies for "sun-following" sensors - at a certain angle, the "pilot" sensor deliberately deviates from the actual angle to compensate.
I was thinking about using solar panels to provide data for decisions about shading in case there is an optimum power setting that strikes a balance between no shading on the one hand and panels directly aimed at the sun on the other. To my mind that requires an assessment of the power available at different angles. Of course it may be possible to do some experiments and arrive at a rule-of-thumb that would be easier to implement.
...R
Robin2:
nebulae:
Every panel costs $500 $)And how much is the expected annual income? - cost on its own is not informative.
You don't need to check for shading very often. Perhaps you could take two panels off-line for a minute or 5 once per hour so they could be used to check shading?
...R
It's a experimental off-grid system. 8)
To take off I need high voltage breaker and LDR is more convenient.
Paul__B:
A few aspects:You clearly do not need extra solar panels to do this, two or three LDRs mounted at top(, centre) and bottom of a representative (mid-array) panel would suffice, or perhaps two or three panels in case of something odd happening to individual panels (birds roosting?).
All tracking algorithms are designed for slow response, so as not to be "tricked" by shadows or clouds.
Yes. You have right. Maybe I also try to use "slow" maximum power point tracking algorithm.
nebulae:
Every panel costs $500 $)
Well, I guess that sums it up. If you are going to invest in the extra panels you are probably better off just using them to produce the extra power to make up for the loss incurred by the shading.
If you can't fix the problem by sensible design of the array in the first place, particularly bearing in mind that you specifically say single axis, the first thing you need to do is assess whether you really have a problem that merits attention. If so, it will be a matter of rather fine adjustment but still simply a matter of modifying the tracking process. To do this there is no need to mess about with extra panels, and using LDRs to test what's shaded and what's not hardly merits contemplation.
At the risk of sounding glib, all you need do is lead or lag the tracking with appropriate programming. What you have going for you is that the sun's path in the sky is entirely predictable for any minute of the year and, by virtue of that, the shadows it casts are equally predictable. Furthermore the changes occur in a regular manner. The geometry to expedite this isn't that hard, and I guess somebody has already done it.