Sun hours

[SunEagle]

Looking good bruce

[AzRoute66]

So, this is all East facing right?

[bcroe]

As the good weather ends, I finally got the new foundation concrete finished. 6 supports for 6 bearings, for a
variable tilt array. The 12 L bolts are in place for above ground attachment. Since I don't want it to emulate
the HOLLYWOOD sign (like the original), a full size 66' X 5' fixture was built to hold the L bolts at exactly the
right location and orientation: straight and square, but sloped to my 10% grade.

They certainly are in a straight line; will be easier to do the final shimming for bearing alignment. That is my
yellow laser on the side. Bruce Roe


Hollywood.JPG




StrButSlope.JPG
LboltFx.JPG

[bcroe]

At this point my net metering KWH reserve for winter is exactly the same as the best of the previous
4 years. Should be plenty, but it always depends on how severe the winter is. Meantime there has
been very slow progress on my southern "solar acre", looks like this today. The input on building a
"percentage above or below MPPT" meter will also be used when this is finished, in theory to set tilt.

Bruce Roe


PVSep17prgr.JPG

[bcroe]

I noticed the panels the other day when I went to check where the bad weather and
tornado was predicted in your area.

Yes, we sent a bunch more rain water down the river this week, brief local flooding. My lot is not rectangular, and the
map shows my installer just lined up on the south fence. Since then I have found the survey pins and plotted out a
pretty accurate E-W line for future work. Its about a 10% grade downhill NE, and everything will be mounted that way.
A before and after reveals a lot of rough brush and shading is now gone. Bruce Roe

[SunEagle]

After 4 years google earth has updated. After 4 years can finally see all my solar panels. Even the short
timed test panels; apparently taken around the end of winter. Bruce Roe

I noticed the panels the other day when I went to check where the bad weather and tornado was predicted in your area.

[bcroe]

After 4 years google earth has updated. After 4 years can finally see all my solar panels. Even the short
timed test panels; apparently taken around the end of winter. Bruce Roe

[bcroe]

By beginning afternoon the WEST system had collected 47 KWH, but the EAST
system had collected only 40 KWH. The difference is the rising sun hitting the WEST system first, completely
clearing the shadows on EAST much later. So WEST is running some 18% ahead for this time. Toward the
end of the day EAST about catches up, which means the same effect is reversed for the setting sun.

It appears that relocating panels can solve most of the discrepancy, without cutting another series of trees.
Bruce Roe

Getting that extra 18% isn't an urgent project, but the sun is out and things are beginning to happen. Last
night I went out in the dark and got the (12 fuse) combiner box ready. Getting close to ordering enough
racking to support 24 panels facing the rising sun, and out of the shade. Some other day, the setting sun.
Then there is the matter of digging a foundation...

Anyway nice to see sun again in this oft cloudy area, been running about 140 KWH for 7 of the last 8 days.
The DC wiring was cool, the AC wiring warms quite a bit after running maxed 8 hours. Another upgrade?
Bruce Roe

[J.P.M.]

Vclipping/Vmp is a tough number to use. The IV curve is really non-linear near the maximum power point, so the voltage change per watt of power "clipped" is probably not constant, nor solely a function of temperature.

Dan's suggestion helps ballpark the maximum energy that could have been produced on a clear day, but I understand that isn't quite what you want. I'm not sure there is a satisfactory answer without taking enough data to build a single diode model of your specific array across enough conditions to be useful. The CEC model implemented in SAM is a good starting point.

FWIW, I'd agree with that. Depending on how the inverter(s) root around looking for an Mpp, V at clipping can be galloping around quite a bit at/if clipping. Most all measurements any backyard fanatic such as Bruce, you, I, etc., can pull off with any reasonable meaning are best done when things are at least semi or quasi stable. FWIW, all my fouling data/stuff is done under as clear conditions as possible and at the time of min. daily incidence angle on the array. That's usually about 19 min. or so after solar noon for my stuff. I apologize if I was less than clear on that point.

[sensij]

Vclipping/Vmp is a tough number to use. The IV curve is really non-linear near the maximum power point, so the voltage change per watt of power "clipped" is probably not constant, nor solely a function of temperature.

Dan's suggestion helps ballpark the maximum energy that could have been produced on a clear day, but I understand that isn't quite what you want. I'm not sure there is a satisfactory answer without taking enough data to build a single diode model of your specific array across enough conditions to be useful. The CEC model implemented in SAM is a good starting point.

[J.P.M.]

Thanks, there will always be some temp error; if its pretty consistent, most will cancel out. Sensors here are pretty
small, TO92 transistor package can be stuck on and insulated on one square inch. If I want to have a simple
read out, I won't be using super accurate IR readings. Bruce Roe

Understood. After being around thermal stuff for 40+yrs., I've learned to not take what I'm told by instruments as gospel. I'd not be surprised if you've found something similar and behave accordingly

Regards,

[bcroe]

You're most welcome.

On sensors, temp. measurement and accuracy: I'd keep in mind that thermosensors are better for use in measuring temperature differences at a point under different conditions rather than as an accurate measurement of panel or cell temperature. (More ) Accurate temp. measurements require low or less thermal resistance between the sensor and the panel than between the sensor and the surrounding environment. Hence, the best readings are from instruments in intimate contact with what's being measured and lots of insulation between the sensor and the external environment. The result of all that is that a panel's local temp. around the well applied sensor is higher than the rest of the panel (if in daylight) because of the back insulation that prevents heat transfer to the environment from the back of the panel, raising the panel temp. and so probably a cause of an erroneous temp. that's not as representative of actual panel temps. Conversely, a poorly or loosely affixed and/or uninsulated sensor will give an inaccurately low reading. Relative error from fixation can be estimated as roughly in proportion to the ratio of thermal resistances between the sensor to panel and sensor to environment.

All that's why the best readings ( or at least not the worst) are usually done in the field w/ IR thermometers when other than relative temp. differences are sought, and they're probably not the best choice in those situations either. Regards,

Thanks, there will always be some temp error; if its pretty consistent, most will cancel out. Sensors here are pretty
small, TO92 transistor package can be stuck on and insulated on one square inch. If I want to have a simple
read out, I won't be using super accurate IR readings. Bruce Roe

[J.P.M.]

Thanks for the input. I think I can do it all at the combiner box, which is next to the ground and directly under
(one end of) the array, minimum copper voltage loss. Some electronic sensors may be attached directly to
the underside of panels, perhaps 4 spread around and averaged as you suggest. I think comparing voltage
out of clipping will give calibration (turn off a couple strings). Then the higher voltage seen with clipping can
be compared to the cell output curve, to get a decent estimate of % clipped. That can be cranked into the
electronics to be automatic. Probably need to do this for seasonal temp shifts to cover a year, to improve on
V/degree calculations.

So I propose comparing Vclipping to Vmp should allow a decent estimate, allowing for the operating temp.
1% not needed, but would like to detect 5% to 15%. With all the rain, don't think dirt is a factor. Bruce

You're most welcome.

On sensors, temp. measurement and accuracy: I'd keep in mind that thermosensors are better for use in measuring temperature differences at a point under different conditions rather than as an accurate measurement of panel or cell temperature. (More ) Accurate temp. measurements require low or less thermal resistance between the sensor and the panel than between the sensor and the surrounding environment. Hence, the best readings are from instruments in intimate contact with what's being measured and lots of insulation between the sensor and the external environment. The result of all that is that a panel's local temp. around the well applied sensor is higher than the rest of the panel (if in daylight) because of the back insulation that prevents heat transfer to the environment from the back of the panel, raising the panel temp. and so probably a cause of an erroneous temp. that's not as representative of actual panel temps. Conversely, a poorly or loosely affixed and/or uninsulated sensor will give an inaccurately low reading. Relative error from fixation can be estimated as roughly in proportion to the ratio of thermal resistances between the sensor to panel and sensor to environment.

All that's why the best readings ( or at least not the worst) are usually done in the field w/ IR thermometers when other than relative temp. differences are sought, and they're probably not the best choice in those situations either.

Dirt, or the lack of it can be a complicating factor, but, more no brainer, a lot of rain will tend to keep things clean(er).

Regards,

[bcroe]

Bruce: FWIW, after measuring both panel instantaneous individual panel temps and quasi simultaneously monitoring array voltages many times, as is needed to estimate array fouling, I've found a pretty good 1st approx. to ave. array temp. change for any string can be found by reading the voltage off my monitor.

To get some measure of Voltage change per deg., I'd measure each of 16 panels w/an IR thermometer at 4 points per panel, then run to the monitor and record string voltages, then run back to the roof and repeat the measurements, all in ~ 12-16 min. around min. incidence angle on very clear days. I repeated this well over 200 times and chose 60 days, 34 in winter, and 26 days the following summer that were tightly grouped and consecutive as much as weather allowed, wishing and knowing that the N==26 was less than perfect, and ran the Voltages and a lot more variables including all weather variables including Horizontal Irradiance recorded at 1 min. intervals and other measurements (example - roof temps. under the array in 16 spots), and ran a normal distribution on all the variables. One result of all that was to get what seems a pretty good way to est. a representative array temp. from array or string voltages.

Bottom line: To SWAG ave. panel temp. from voltage: Note the voltage at any two times and divide that voltage diff. by (the published Vmpp change per deg. C. times the number of panels in the string). Long story how I got the measurement method right, and there's a lot more to it, particularly for extrapolating the voltage at an ave. array temp. of 25 C. to guesstimate the voltage drop up to the inverter, but that method is the basis for it and, after all the hoopla described above and some tweaking, it has shown to be quite reliable over many measurements.

More no -brainer: (Current per string)/(Impp) is a pretty good 1st approx. to P.O.A. irradiance. Just multiply that ratio (string current/Impp) by 1,000 to get Watts /m^2 P.O.A. irradiance on the array or string. The result compares quite favorably with measured GHI corrected to P.O.A. using the HDKR algorithm, and a pretty good way to treat your array as a pyranometer.

Take what you want of the above, scrap the rest.

Thanks for the input. I think I can do it all at the combiner box, which is next to the ground and directly under
(one end of) the array, minimum copper voltage loss. Some electronic sensors may be attached directly to
the underside of panels, perhaps 4 spread around and averaged as you suggest. I think comparing voltage
out of clipping will give calibration (turn off a couple strings). Then the higher voltage seen with clipping can
be compared to the cell output curve, to get a decent estimate of % clipped. That can be cranked into the
electronics to be automatic. Probably need to do this for seasonal temp shifts to cover a year, to improve on
V/degree calculations.

So I propose comparing Vclipping to Vmp should allow a decent estimate, allowing for the operating temp.
1% not needed, but would like to detect 5% to 15%. With all the rain, don't think dirt is a factor. Bruce

[bcroe]

Can't you eyeball that already by just eyeballing the power output curve, and
extrapolating where the curve would go if it weren't clipped?

Dan, that isn't the right approach. You may remember (or can go back over many curves here) that the array
doesn't run the classic SW desert curve at all. And the reason for pursuing this is to CHANGE the curve to
something where the potential power curve is just the minimum over the clipping level, at the maximum number
of hours. In any case plotting any of these curves is quite a time consuming all day project, highly dependent
on a mostly clear sky.

But I think a temp-MPPT voltage reading can give me an instantaneous answer. I may be out there cranking
the inclination and watching the meter. Bruce