System Performance degradation

[Brianellul]

Hi All

Last week I was chatting with my supplier and he pointed out that my system in not performing well. To be honest I was aware that the KHW numbers went down but just convinced myself that this year, we have more clouds and the Summer is hotter (which is true after all). However my supplier insisted that the system is not performing well and was asked to make somechecks for him.

The table below is the average daily production (KHW) for 2011 and 2012, as you can notice 2012 is lower for all months. The system has been in operation in December 2010.

Month 2011 - 2012
Jan 5.22 - 4.10
Feb 6.42 - 5.51
Mar 8.20 - 6.87
Apr 9.93 - 7.80
May 10.84 - 8.83
June 9.94 - 8.71
July 9.80
Aug 9.48
Sep 8.05
Oct 6.03
Nov 4.14
Dec 4.11

My system is grid tie, nine IBC 230W solar panels for a total of 2070W (configured as one string) and the grid tie inverter is the transformerless SMA SB 2100TL.

I Checked the Short Circuit current & Open Voltage for the nine panels at 11:00 local time (Panels should peak at about noon / 13:00...):
Today the thermometer reached 33 degrees Celcuis with a heat stress of about 40 degrees Celcuis (according to Met Office).

Panel 1 - 25.5 Voc / 6 Asc
Panel 2 - 27.5 Voc / 5.97 Asc
Panel 3 - 26.8 Voc / 5.92 Asc
Panel 4 - 29.0 Voc / 5.92 Asc
Panel 5 - 26.1 Voc / 5.81 Asc
Panel 6 - 27.8 Voc / 5.83 Asc
Panel 7 - 32.4 Voc / 6.39 Asc
Panel 8 - 32.2 Voc / 6.37 Asc
Panel 9 - 31.9 Voc / 6.3 Asc

The readings look fine at least in my opinion. Panels 1-6 are mounted in parallel to the building and are not facing truly South, in fact their output before noon is slightly less than the other three panels. Panels 7-9 are facing South.

I took the Open Circuit Voltage of all nine panels in series (261v) while the Short circuit current is 6.78Amps. These reading were taken on the roof.

I also took the Open circuit voltage of all nine panels in series (262v) while the Short circuit current is 7Amps at the Inverter input. (This was taken after the roof measurements in a time span of 10mins and there is a length of about 12 meters 10mm copper cable). From this I can conclude that the wiring is OK and there is no voltage drop from the panels to the inverter?

Once I connected the Inverter, as expected, the voltage went down to 200v while the current 5.42, the inverter exporting 989W. (All these last 3 readings taken directly from the inverter display)

At 16:00 I took more readings: 264Voc / 7.67Asc while the inverter was exporting only 1047W, Vpv 189. The inverter has been all day on MPP, no errors/ warnings and the max exported wattage has never exceeded 1200W.

I'm tempted to point my finger at the inverter MPPT... what's your opinion and if anyone has any recomendations for me to follow.

Regards
Brian

[ggunn]

I also took the Open circuit voltage of all nine panels in series (262v) while the Short circuit current is 7Amps at the Inverter input. (This was taken after the roof measurements in a time span of 10mins and there is a length of about 12 meters 10mm copper cable). From this I can conclude that the wiring is OK and there is no voltage drop from the panels to the inverter?

If there is no current flowing there can be no voltage drop, so it wouldn't show up in a Voc measurement.

[niel]

if the voc was measured at the pvs (isolated that is not connected to anything else) themselves with full sunshine even at 11am then i would say the pvs are out of spec. you should ask your installer to verify what you've found and if this is the case then get your pvs replaced under warranty.

the isc can vary with conditions, but voc should be in the ballpark and you are showing it way down in all cases.

[ggunn]

if the voc was measured at the pvs (isolated that is not connected to anything else) themselves with full sunshine even at 11am then i would say the pvs are out of spec. you should ask your installer to verify what you've found and if this is the case then get your pvs replaced under warranty.

the isc can vary with conditions, but voc should be in the ballpark and you are showing it way down in all cases.

I agree, mostly.

The Voc for the module is 36.8V. The ambient temp was 33 degrees, and if his modules are flush mounted they could be as much as 35 degrees C hotter. The temperature coefficient for Voc is -121mV/degree C. If his cell temp is 68 degrees C, he should expect Voc to be (68-25)(.121) = 5.2V down from 36.8V, or about 31.6V. Three of his modules showed Voc of about that, but the other six were substantially lower.

Statistically speaking, it looks like a bimodal distribution with three of them grouped around 32V and the other six around 27V; if his problem were something random, I would not expect that. Also, the group of six was all together (Panels 1-6) as was the group of three (Panels 7-9). That all looks a little odd and implies to me that there might be something else going on. What that could be, though, I dunno. Maybe he should look at the module nameplates and see if all the modules are the ones he thinks he has.

[inetdog]

I agree, mostly.

The Voc for the module is 36.8V. The ambient temp was 33 degrees, and if his modules are flush mounted they could be as much as 35 degrees C hotter. The temperature coefficient for Voc is -121mV/degree C. If his cell temp is 68 degrees C, he should expect Voc to be (68-25)(.121) = 5.2V down from 36.8V, or about 31.6V. Three of his modules showed Voc of about that, but the other six were substantially lower.

Statistically speaking, it looks like a bimodal distribution with three of them grouped around 32V and the other six around 27V; if his problem were something random, I would not expect that. Also, the group of six was all together (Panels 1-6) as was the group of three (Panels 7-9). That all looks a little odd and implies to me that there might be something else going on. What that could be, though, I dunno. Maybe he should look at the module nameplates and see if all the modules are the ones he thinks he has.

If the Voc of the panels (rated) is 36.8 and the maximum measured Isc is around 6.4A, I have a hard time getting a 230 Watt panel out of that given that Vmp and Imp will have to both be lower.
For the low 6 panels it looks even worse, but they are not getting perpendicular insolation.

Also, the one measurement which I see only once, as measured by the inverter itself is Imp. We can indirectly compute a value for the other time by dividing the exported power by the displayed Vmp and then fudging for assumed inverter efficiency. 1047w divided by 189v is about 5.5A. That seems consistent with the measured Isc for the string.

Since three panels are close to South and six panels are not as close, the differences between the sets seems as expected, but the design using one MPPT inverter for both sets in a series string is questionable.

[ggunn]

If the Voc of the panels (rated) is 36.8 and the maximum measured Isc is around 6.4A, I have a hard time getting a 230 Watt panel out of that given that Vmp and Imp will have to both be lower.
For the low 6 panels it looks even worse, but they are not getting perpendicular insolation.

Also, the one measurement which I see only once, as measured by the inverter itself is Imp. We can indirectly compute a value for the other time by dividing the exported power by the displayed Vmp and then fudging for assumed inverter efficiency. 1047w divided by 189v is about 5.5A. That seems consistent with the measured Isc for the string.

Since three panels are close to South and six panels are not as close, the differences between the sets seems as expected, but the design using one MPPT inverter for both sets in a series string is questionable.

From the data sheet (link in post #1): Vmp = 29.6V, Imp = 7.78A. (29.6V)(7.78A) = 230.3W

Like modules won't have different Voc values simply because they have different orientations. Maybe the low reading ones were hotter? But they would have to be at 106 degrees C to depress Voc from 36.8V to 27V @ -121mV/degree C. Curiouser and curiouser.

I agree that having modules in a string pointed in different directions is bad design; the string current will always be limited to that of the worst performing module in the string. That could explain a lot, the screwy looking measurements notwithstanding.

[Brianellul]

If there is no current flowing there can be no voltage drop, so it wouldn't show up in a Voc measurement.

You're right! Without any current flowing there will be no voltage drop. I should have worded differently... basically I concluded that there are minimal losses in the 12meter run.

Thanks

[Brianellul]

if the voc was measured at the pvs (isolated that is not connected to anything else) themselves with full sunshine even at 11am then i would say the pvs are out of spec. you should ask your installer to verify what you've found and if this is the case then get your pvs replaced under warranty.

the isc can vary with conditions, but voc should be in the ballpark and you are showing it way down in all cases.

Neil,thanks for your feedback! I was not aware the voc should not vary with conditions. I was under the impression that both the voc and isc will vary, both reaching their maximum when the sun is perpendicular to the panels... Is this what you're telling me?

[Brianellul]

I agree, mostly.

The Voc for the module is 36.8V. The ambient temp was 33 degrees, and if his modules are flush mounted they could be as much as 35 degrees C hotter. The temperature coefficient for Voc is -121mV/degree C. If his cell temp is 68 degrees C, he should expect Voc to be (68-25)(.121) = 5.2V down from 36.8V, or about 31.6V. Three of his modules showed Voc of about that, but the other six were substantially lower.

Statistically speaking, it looks like a bimodal distribution with three of them grouped around 32V and the other six around 27V; if his problem were something random, I would not expect that. Also, the group of six was all together (Panels 1-6) as was the group of three (Panels 7-9). That all looks a little odd and implies to me that there might be something else going on. What that could be, though, I dunno. Maybe he should look at the module nameplates and see if all the modules are the ones he thinks he has.

In fact you're right, Panels 1-6 are grouped and will reach their maximum output 30-45mins after panels 7-9 (that's why the difference in output for the panels). Yes the panels are identical, I've rechecked the nameplates.

Something which I should consider after reading all your comments is that I should allign ALL 9 panels. Panels 1-6 are totally fixed and can't be moved/tilted, however panels 6-9 can be tilted to match the first 6 panels, here I'm talking just a couple of degrees. In this way all panels will peak at the same time.

Having said all this, last year (i.e. installation year), I used to see peaks > 2KW output from the inverter and it's only this year which I'm noticing a decrease in output.

[Brianellul]

Guys, thanks to all for your feedback. I'm currently still waiting for the installer to comeback with feedback from his supplier... will let you allknow what's the outcome.

Regards
Brian