Selling home, solar inspection issue with VOC measurement.

[bcroe]

To get more accurate voltage measurements, the usual method is remote sensing, a pair of wires directly from
the point of interest to the V meter. Suitable protection needed in this case, could be a couple 1A fuses. That
still doesn't account for the panel leads, some 60' for a string here. Individual panels could go direct for the
bypass connections. Bruce Roe

[J.P.M.]

right and OP problem was low Voc so either temp was unusually high (for me) or irradiance low.

Or something was not being measured correctly or accurately or accurately, or something was wrong with the system.


how would you explain right side of IV curves where they intersect I = 0 level? They clearly show irradiance influence on Voc.

J.P.M. added:

The open circuit Voltage does increase as f(irradiance), but the increase is logarithmic. What everyone (or at least me) is trying to tell you is not that Irradiance does not increase voltage - it does, but that the effects of temperature on voltage are so much larger than the effects of temp. on Voltage that the effects of temp. on Voltage mask most any theoretical but still real effects that increases in irradiance will have on Voc. Basically, any increase in Voltage due to irradiance increases are drowned out or masked by temp. effects, particularly under operating conditions that normally have higher POA irradiance levels to work with as can be seen from the more constant and very steep slope of the locus of Mpp's as irradiance increases and temp. is held constant.
At those higher irradiance levels, temp. change effects on voltage are something like an order of maginude or two more than irradiance effects and so, he irradiance effects can usually be ignored. I believe I might be able to see some of the irradiance effects in the data I described, with a slight difference in average Voltage for the summer data (higher POA irradiance) vs. winter data (lower POA irradiance). But I can't claim any accuracy on that, and given the shape and slope of the likely locus of MPP's at those ave. POA's (830 W/m^2 winter, 965 W/m^2summer) I wouldn't expect to. Point is, for me anyway, the irradiance/Voltage relationship is hard to measure and can usually be rolled in with the Temp. effects on voltage with little, if any loss in accuracy.

BTW, going completely scientific about this when measuring cell temperature what should be considered cell temperature? I think back of the cell is approximation down while front- approximation up as the actual element is behind layer of glass at least. I understand that energy comes from the front and then dissipates in all directions and some of it gets converted cooling cell a bit I just wonder how big those effects are- fraction of degrees or few whole degrees? Did you come across data where someone measured those temps using infrared thermometers? Those should provide some insight as it would be relatively hard to measure temp of the front of the cell with some sort of contact sensor without influencing it if we talk about fraction of degree here - the temp probe would be in the path of the incoming energy flow shading the cell it is intended to measure.

To your other points: See Duffie & Beckman for discussions of cell temp. vs. plate temps., or see the Sandia method for estimating cell temps. There is a paper that gives a somewhat detailed description that's pretty close to a lot of what I did to get an energy balance on my array. It's on the net. Look for :"A thermal model for photovoltaic panels under varying atmospheric conditions", Armstrong & Hurley, Applied Thermal engineering, vol. 30 (2010), pp. 1488 -1495.

I used the Sandia method to estimate cell temps. As I described in a prior post, I was actually measuring the back plate temp. of the panels, and adjusting that temp. to estimate the cell temp. as the Sandia method suggests. I used an IR thermometer that I did a home calibration method on, using boiling water, an ice bath and my body temp. as 3 calibration points. I adjusted the emissivity for the back plate using data from engineering literature about the emissivity of tedlar (==0.90).

What you are writing and asking about deals with energy transfer - and is the meat and business of how to account for what happens to the energy that flows to and from an object, be it a PV array or anything else. Read the sources I listed and check out a heat transfer text. Look up "energy balance".

[max2k]

See Duffie & Beckman for discussions of cell temp. vs. plate temps., or see the Sandia method for estimating cell temps. There is a paper that gives a somewhat detailed description that's pretty close to a lot of what I did to get an energy balance on my array. It's on the net. Look for :"A thermal model for photovoltaic panels under varying atmospheric conditions", Armstrong & Hurley, Applied Thermal engineering, vol. 30 (2010), pp. 1488 -1495.

I used the Sandia method to estimate cell temps. As I described in a prior post, I was actually measuring the back plate temp. of the panels, and adjusting that temp. to estimate the cell temp. as the Sandia method suggests. I used an IR thermometer that I did a home calibration method on, using boiling water, an ice bath and my body temp. as 3 calibration points. I adjusted the emissivity for the back plate using data from engineering literature about the emissivity of tedlar (==0.90).

What you are writing and asking about deals with energy transfer - and is the meat and business of how to account for what happens to the energy that flows to and from an object, be it a PV array or anything else. Read the sources I listed and check out a heat transfer text. Look up "energy balance".

I think I see where this all coming from- I consider the tail of the process where irradiance low while you and sensij- higher levels of it. Since Voc depends logarithmically from irradiance and linearly from temperature their net effect is mostly controlled by irradiance at low level (below 300 W/m2) and temperature - above that level. These 2 components have opposite signs so they somewhat compensate each other and there must be a point on irradiance curve (if we let cell temp rise with increasing irradiance) where they fully compensate each other: not high enough to flatten out on logarithm part and not yet high enough on linear temperature part. So this discussion is about where OP array was - to the left of that point or to the right? I say to the left as 300+ irradiance levels and cell temperatures above 40C at 8:00 AM sounds stretchy to me but I'm making assumptions here, may be it was that hot already.

[J.P.M.]

I think I see where this all coming from- I consider the tail of the process where irradiance low while you and sensij- higher levels of it. Since Voc depends logarithmically from irradiance and linearly from temperature their net effect is mostly controlled by irradiance at low level (below 300 W/m2) and temperature - above that level. These 2 components have opposite signs so they somewhat compensate each other and there must be a point on irradiance curve (if we let cell temp rise with increasing irradiance) where they fully compensate each other: not high enough to flatten out on logarithm part and not yet high enough on linear temperature part. So this discussion is about where OP array was - to the left of that point or to the right? I say to the left as 300+ irradiance levels and cell temperatures above 40C at 8:00 AM sounds stretchy to me but I'm making assumptions here, may be it was that hot already.

I'd suggest checking out the sources provided here. I'm not going to continue along this line. I feel like I'm either talking to a wall or bring played for a fool.

[max2k]

I'd suggest checking out the sources provided here. I'm not going to continue along this line. I feel like I'm either talking to a wall or bring played for a fool.

Same here and as I said we can just leave it at it. I definitely have no intention to waste your or sensij time but I do have problem going into model analysis ignoring direct data in form of IV curves.

I think the simple experiment we can try is to take few evenly time spaced measurements of Voc in the morning and if what I suspect is true when irradiance will be increasing also heating up cells that curve would have max value where irradiance influence will be overcome by temperature. If the temperature is the only prevailing factor through entire range of irradiance levels the curve won't have max and will smoothly decrease. I doubt latter is possible as in full dark Voc is definitely 0 so it has to get to non 0 value somehow and that is the 'rising' part of the function. I'll do it to improve my own understanding, just need to wake up early enough . My array is just dumb string so I don't have to climb roof, just play with disconnects in the right order. Irradiance can be estimated from the power output curve so when superimposed those should answer this dilemma for me.

[sensij]

Same here and as I said we can just leave it at it. I definitely have no intention to waste your or sensij time but I do have problem going into model analysis ignoring direct data in form of IV curves.

I think the simple experiment we can try is to take few evenly time spaced measurements of Voc in the morning and if what I suspect is true when irradiance will be increasing also heating up cells that curve would have max value where irradiance influence will be overcome by temperature. If the temperature is the only prevailing factor through entire range of irradiance levels the curve won't have max and will smoothly decrease. I doubt latter is possible as in full dark Voc is definitely 0 so it has to get to non 0 value somehow and that is the 'rising' part of the function. I'll do it to improve my own understanding, just need to wake up early enough . My array is just dumb string so I don't have to climb roof, just play with disconnects in the right order.

What direct data do you think is being ignored? The IV curves you keep referencing show that the irradiance effect is likely to be less than the temperature effect at the conditions we are talking about. You aren't wrong that there is a transition point between irradiance and temperature early in the morning, but you are mis-identifying the point at which that transition occurs.

Here is more real data. This is mppt voltage and current, not voc and isc, but is directionally relevant.

This is from my most recent system, on 7/23/17. It didn't get especially hot that day, so the temperatures are probably a bit lower than they would be in the OP's case, but it is one of the rare days recently that there hasn't been much marine layer around here. The peak mppt voltage occurred around 7:20 am, and by 8:00 am, the temperature effect is clearly dominating the irradiance effect.
2017-07-23.JPG



Here is data from a hotter day, on my older system (06/19/2016). Obviously, the day is slightly longer so the peak mppt voltage shifts slightly earlier. You can see how much more steeply voltage drops, because of how much hotter it is getting that day.
2017-06-19.JPG


Next to my old array, I had an irradiance sensor mounted. At 8 am on 6/21, it reported a GHI of 374 W/m2. On 7/23/16, it reported 330 W/m2. Both of those numbers would need to be converted to plane of array irradiance to be useful for any performance modeling, but hopefully that gives a sense that the irradiance at that time in the morning is somewhat brighter than you seem to be thinking.

[max2k]

What direct data do you think is being ignored? The IV curves you keep referencing show that the irradiance effect is likely to be less than the temperature effect at the conditions we are talking about. You aren't wrong that there is a transition point between irradiance and temperature early in the morning, but you are mis-identifying the point at which that transition occurs.

Here is more real data. This is mppt voltage and current, not voc and isc, but is directionally relevant.

This is from my most recent system, on 7/23/17. It didn't get especially hot that day, so the temperatures are probably a bit lower than they would be in the OP's case, but it is one of the rare days recently that there hasn't been much marine layer around here. The peak mppt voltage occurred around 7:20 am, and by 8:00 am, the temperature effect is clearly dominating the irradiance effect. 2017-07-23.JPG




Here is data from a hotter day, on my older system (06/19/2016). Obviously, the day is slightly longer so the peak mppt voltage shifts slightly earlier. You can see how much more steeply voltage drops, because of how much hotter it is getting that day. 2017-06-19.JPG



Next to my old array, I had an irradiance sensor mounted. At 8 am on 6/21, it reported a GHI of 374 W/m2. On 7/23/16, it reported 330 W/m2. Both of those numbers would need to be converted to plane of array irradiance to be useful for any performance modeling, but hopefully that gives a sense that the irradiance at that time in the morning is somewhat brighter than you seem to be thinking.

now that looks like something. I'm fairly sure your assumption Vmp behaves in synch with Voc is correct but let me at least try to measure one of my strings or even both and report back here. Most likely it will confirm what both you and JPM were saying all along that irradiance effect was buried in temp dependency by the time OP Voc measurement was taken.

I'm going to manually capture measured Voc, Vmp and inverter reported power output with 10 min intervals between 6 and 9 AM as even I believe by 9 AM Voc will pass its max.

On slightly unrelated topic if we consider function y = lg(x) - x it has max at x = 1 and at that point both linear and log components contribute equal but opposite values. So we can say we measure 'true' Voc only at those max points .

[sensij]

One more look at it, this time using NREL's System Advisor Model, set to use the CEC performance model I described earlier.

I "calibrated" SAM by looking at the OP's PVOutput data from 2016. Wunderground says 7/23/16 had a high of 106 deg F in Redding, so probably similar to the day the Voc measurement was recently performed. Scanning through the TMY3 weather file for Redding, the 7/24 entry hits a peak ambient temperature of 41.7 C (107 deg F), so that should be a reasonably good match.

Loading in the OP's panels, inverter, and orientation, taking a stab at the other mounting related inputs, and adjusting the wiring and other loss factors to match the PVOutput results, the following chart can be produced: shastalake.JPG





The match in the power curve is "close enough" for me.

The afternoon wiggle in the SAM data comes from the TMY3 file knocking the wind down from 2.6-3.1 m/s to 0 for one of the afternoon hourly points. I'm not sure if that is intentional, but it creates a spike in cell temp that lowers the voltage for that hourly interval. The OP's system appears to have some evening shade or an obstructed view of the horizon, leading to some divergence from the model after around 5:30 pm. But that isn't really what we are looking at here.

The Voc at 8 am in this model is about 373 V, really close to the 378 V reported by the OP in the first post. If nothing else, I hope this further cements the idea that the voltage that was reported looks perfectly healthy.

Speculating on what "contributes" to the voltage reduction relative to the Voc at STC, here are a few outputs from the model at 8 am:

POA Irradiance is around 240 W/m2.

Tcell is about 37 deg C.

Voc and Vmp don't follow each other as closely as I had thought in the morning.

max2k will see that the Voc is continuing to rise until after 9 (even though the mpp voltage started falling earlier), so there is support here for the idea that irradiance effects on Voc are contributing more strongly to the voltage than I had allowed earlier. Really digging that out from inside the model will take more work (I have a Python version of SAM that I wrote that allows me to manipulate the variables as needed, but there is some hard-coded stuff in there for my own system that would need to be re-written for the OP's system to do this).

I've attached the SAM project file (with a *.pdf extension appended to make the forum server happy) for anyone who might want to take a look.

[J.P.M.]

One more look at it, this time using NREL's System Advisor Model, set to use the CEC performance model I described earlier.

I "calibrated" SAM by looking at the OP's PVOutput data from 2016. Wunderground says 7/23/16 had a high of 106 deg F in Redding, so probably similar to the day the Voc measurement was recently performed. Scanning through the TMY3 weather file for Redding, the 7/24 entry hits a peak ambient temperature of 41.7 C (107 deg F), so that should be a reasonably good match.

Loading in the OP's panels, inverter, and orientation, taking a stab at the other mounting related inputs, and adjusting the wiring and other loss factors to match the PVOutput results, the following chart can be produced: shastalake.JPG





The match in the power curve is "close enough" for me.

The afternoon wiggle in the SAM data comes from the TMY3 file knocking the wind down from 2.6-3.1 m/s to 0 for one of the afternoon hourly points. I'm not sure if that is intentional, but it creates a spike in cell temp that lowers the voltage for that hourly interval. The OP's system appears to have some evening shade or an obstructed view of the horizon, leading to some divergence from the model after around 5:30 pm. But that isn't really what we are looking at here.

The Voc at 8 am in this model is about 373 V, really close to the 378 V reported by the OP in the first post. If nothing else, I hope this further cements the idea that the voltage that was reported looks perfectly healthy.

Speculating on what "contributes" to the voltage reduction relative to the Voc at STC, here are a few outputs from the model at 8 am:

POA Irradiance is around 240 W/m2.

Tcell is about 37 deg C.

Voc and Vmp don't follow each other as closely as I had thought in the morning.

max2k will see that the Voc is continuing to rise until after 9 (even though the mpp voltage started falling earlier), so there is support here for the idea that irradiance effects on Voc are contributing more strongly to the voltage than I had allowed earlier. Really digging that out from inside the model will take more work (I have a Python version of SAM that I wrote that allows me to manipulate the variables as needed, but there is some hard-coded stuff in there for my own system that would need to be re-written for the OP's system to do this).

I've attached the SAM project file (with a *.pdf extension appended to make the forum server happy) for anyone who might want to take a look.

FWIW, nicely done. Looks like pretty reasonable agreement. Question: Graph times - looks like they're PST - correct ?

Thanx,

J.P.M.

[sensij]

FWIW, nicely done. Looks like pretty reasonable agreement. Question: Graph times - looks like they're PST - correct ?

Thanx,

J.P.M.

Yes.. the PVoutput data set the scale, OP's system reports every 5 minutes, so timing could be off plus/minus that amount if the data are leading, lagging, or instantaneous.

SAM data were adjusted by 1 hour for Savings time, and another 30 min to account for the 30 min sun position calculation used to represent each hour. (In other words, when SAM reports data for the 8am hour, it uses sun position at 8:30). I *think* TMY files are constructed the same way (the agreement suggests so), but it has been a while since I looked at that.

[J.P.M.]

Yes.. the PVoutput data set the scale, OP's system reports every 5 minutes, so timing could be off plus/minus that amount if the data are leading, lagging, or instanteous.

SAM data were adjusted by 1 hour for Savings time, and another 30 min to account for the 30 min sun position calculation used to represent each hour. (In other words, when SAM reports data for the 8am hour, it uses sun position at 8:30). I *think* TMY files are constructed the same way (the agreement suggests so), but it has been a while since I looked at that.

Thank you.

[jschner]

Holy smokes you guys are good. Sensj it looks like you nailed it perfectly with that graph. And yes there is some shading as I have a hill on the west side with a few trees on it that I left so I did not have to look at my neighbor.s house. It pretty much shades the west side string and not the east side string.

For an update, I think my call bugged the inspector to go take another look and this time he took an "ambient light" meter? Anyway, he did not call me directly but called the buyers and told them it looks like this system is working with in spec and there was nothing wrong. I got a call from my agent and they agreed to drop the costs of checking the system. So we are all good now. Thanks all for the suggestions and the great amount of data. It all helped and I learned a lot as well.

The house I'm buying in Arizona has solar on it too, fully owned, not a lease, and I believe it will be grandfathered into the old program. I need to double check on that. The current owner does not know anything about it but it has 33 panels all pointed South and a Sunnyboy 7000US on the side of the home put on about 2015 I think. Looks like I will be back in the solar business again.

[AzRoute66]

Hooray. Solarpaneltalk 1 Careless adversarial inspectors 0

[J.P.M.]

Holy smokes you guys are good. Sensj it looks like you nailed it perfectly with that graph. And yes there is some shading as I have a hill on the west side with a few trees on it that I left so I did not have to look at my neighbor.s house. It pretty much shades the west side string and not the east side string.

For an update, I think my call bugged the inspector to go take another look and this time he took an "ambient light" meter? Anyway, he did not call me directly but called the buyers and told them it looks like this system is working with in spec and there was nothing wrong. I got a call from my agent and they agreed to drop the costs of checking the system. So we are all good now. Thanks all for the suggestions and the great amount of data. It all helped and I learned a lot as well.

The house I'm buying in Arizona has solar on it too, fully owned, not a lease, and I believe it will be grandfathered into the old program. I need to double check on that. The current owner does not know anything about it but it has 33 panels all pointed South and a Sunnyboy 7000US on the side of the home put on about 2015 I think. Looks like I will be back in the solar business again.

Sounds like it worked out well for you. Glad this place helped. Sensij gets a tip of the hat on this one. I think "ambient light meter" usually refers to photographic applications. Sounds like B.S. to me, but no matter now. I would, however, suggest getting all of the festivities in writing in some way.

[sensij]

Sounds like it worked out well for you. Glad this place helped.

None of us is as good as all of us.