Need help diagnosing (and optimizing) solar installation

Keithjwaz

We have installed two solar system in the Caribbean at our two small villas in St. Lucia, West Indies. One system is 5k with a total of 11 450 watt (approx.) panels. The other system is 7k with 15 panels. The energy usage (demand) at the smaller villa runs about 550 kWh per month and the usage at the other villa runs about 850 kWh per month, so there is about a 60% difference in terms of energy usage between the two little villas. 

The orientation of the panels for both villas is basically an eastern one as the shed roofs we have angles from west to east at about 18 degrees pitch. The roof is metal so we do get some heat here.

We’ve had all kinds of unfortunate issues with our solar installer (everything from critical employees leaving the company to delays and mistakes in filling out paperwork and applications with the local power company to lack of understanding on how to set up the data logging features of our Solis inverters so that we can even track our true performance). After 5 months of having the panels installed, our solar installer had still not been able to get our two systems connected with the local power company so that we can get proper credit for the energy our two modest power plants create.

Recently, we took matters into our own hands and set up the data loggers with the inverters so that we could get daily data feeds that are sent to the cloud.

We’ve suspected all along that the eastern exposure of our panels and the specific configurations we have are such that our systems are underpowered, especially the 7k one, The data that we are receiving has shown this to be the case. Here’s what we are seeing:

• The 5k system is achieving about 90% of its target. This system we are not too concerned about.
• The 7k system is only achieving 70% of its target (both systems have the same basic orientation and environmental conditions).
• We notice that our peak time for generating the most kWh is in the morning between about 10 am to 1 pm. After this, the power generation really drops down, we see this happen each day of the week. I could post data charts of this performance if that were helpful.

Before the systems were installed, we discussed the issues about having the proper exposure and tilt angles for the panels but the installer just ignored our thoughts in this matter, assuring us that this wouldn't really matter due to our extensive sun exposure in the Caribbean. This didn’t seem right to me but we went along with this approach, 

Given all of the problems that we’ve had with the installer, we think we were mislead and that our systems are not set up as best they could be. When we presented the issues about performance to the installer, his response was that the systems aren’t performing as well as he projected because of shading and leaves, pollen, and debris getting on the panels. (After the install, we cut back trees in the area and the panels have been cleaned.) The panels are also very new and the underperforming villa has been that way since the beginning. 

We’re trying to get this right because electricity rates on our island are sone of the highest found anywhere—every kWh generated really counts. 

So our questions are:

• Are we correct in our assumptions about the exposure of the panels? In other words, isn’t this a design flaw by the installer?
  
• In looking at why the 5k system performs okay but the 7 k system does not, isn’t this just a matter that the 7k system doesn’t have enough panels? After all, the 7 k system only has 4 more panels than the 5 k system. How could the design of one system be just a little larger than the other yet ut is expected to generate so much more energy (550 vs. 850 kWh)?

BB.

Welcome to the forum Keith,
https://pvwatts.nrel.gov/pvwatts.php
Having more information about your system would be helpful... Just for interest, I did a model of your system for Martinque. 18 degree tilt, faciing east, 7 kWatts of panels, and assuming 52% overall system efficiency (assuming flooded cell lead acid battery bank):

Month	Solar Radiation ( kWh / m2 / day )	AC Energy ( kWh )
January	4.33	449
February	5.09	479
March	5.41	557
April	5.72	568
May	5.71	590
June	5.54	552
July	5.59	575
August	5.79	592
September	5.52	546
October	4.89	501
November	4.43	442
December	4.33	450
Annual	5.20	6,301

The above numbers may be 10% low (I have not used PV Watts much lately and they tend to "hide" some of their calculations--But close enough to talk about your harvest).

Not a "lot" of sun (afternoon clouds and/or thunderstorms common?). If you were using AGM or Li Ion batteries, would get "better" numbers (more efficient batteries).... And if were direct connected to the grid in "Grid Tie Mode" and only using the batteries for backup, numbers could be 50% higher...

For a "full time" off grid system, would be suggesting that you use around 50% to 65% of predicted harvest (to allow for stormy weather, larger loads on certain day for washer/drier/cooking/water pumping/working around the home, etc..)...

Also knowing what charge controller (or inverter-solar charger) you are using and the array specifications (Vmp/Imp, how many panels in series and how many parallel strings) could help... You are welcome to post links to the hardware specifications/manuals.

For example, say you have 15 panels in your array configured for 5x panels in series and 3x parallel strings... If you had a bad panel or electrical connection, you could lose 1/3rd of the array current.

Simple tests for this would be: A) if you have a combiner box (3 breakers, one for each series string) you could turn off one breaker at a time and see if the current drops by 1/3rd (each string sharing current).

You could also use an AC+DC Current Clamp Meter (DMM--Digital Mulit-meter) to measure the current in each parallel string (if you don't have a combiner box with breakers).

Also, knowing the exact panel ratings and solar charge controller specifications... More or less, the solar panel Imp current is based on amount of sun hitting the panels. One would expect at/near noon on a clear day, the panels should be producing 50% to 100% of their Imp rated current.

If you have a bad cell/panel, you would see a Vmp-array voltage drop... Instead of seeing Vmp-std*number of panels in series*Vmp*0.8 hot panel derating... You would see much lower array voltage (lower by 12 volts to 30/40/whatever volts).

Another confounding issue is that (I presume) this is a battery based off grid system... The charge controller will only draw as much power from the solar panels as the batteries need to charge. Too small of battery bank, battery bank "issues", too small of charge controller to battery bank connections causing high voltage drop, will also "affect" the solar array "raw numbers" (current and voltage).

If you pointed the array due south, 14 degree tilt, you would get around 10% more hours of sun per day... Not a lot--But with solar, any gain in harvest, savings in load draw, etc. helps.

If you have a battery based system, understanding your battery State of Charge/Charging voltage & current during parts of the day would be helpful too.. A battery bank that is >~80% State of Charge is around the Charging Voltage Set Point and starts to self limit it acceptance of charging current... As the bank approaches 100% SoC, the charging current tails off to ~1% or less (this is for Lead Acid Batteries--Li Ion and other chemistry batteries will be different).

I typically look at "daily" power harvest and usage... Since the battery bank is (usually) only large enough to store a couple days worth of solar harvest and supply loads (on "dark days").

An example of daily harvest for February (math, my first guess at your system):

• 7,000 Watt array * 0.52 off grid FLA battery system eff * 5.09 hours of sun per day = 18,528 Watt*Hours per day = 18.5 kWH per day

And if we use 65% "off grid solar fudge factor" for daily energy usage (average predicted harvest):

• 18,528 WH per February day * 0.65 fudge factor = 12,043 WH suggested Average Loads for February (with 7,000 Watt array in Feb)

If you want to look closer at your system's performance for debugging/better understanding, an AC+DC Current Clamp Meter is very handy. Just zero the meter (for DC usage) and clip the clamp around one wire--And read the current.

Note there also AC only clamp meters... They work very well--But only for AC... Not DC CURRENT (note that an AC clamp meter may measure AC voltages, but will not measure DC current). A couple of examples of AC+DC clamp meters / DMMs:

https://www.amazon.com/UNI-T-Digital-Handheld-Resistance-Capacitance/dp/B0188WD1NE (inexpensive--Good enough for most/smaller DC powered systems)
https://www.amazon.com/Auto-Ranging-Resistance-Klein-Tools-CL800/dp/B019CY4FB4 (mid price meter)

Your thoughts/corrections/additional information?

-Bill

Keithjwaz

Hi Bill,

Thanks much for taking the time to leave such a detailed post with your data and suggestions. I’ll look forward to reviewing your suggestions in more depth and taking some readings when we get a chance. 

For now, here is some more information about our installations and equipment with some further thoughts about our systems. I’m not certain about the actual wiring configuration if the panels but I can certainly look into that aspect. 

I’ll hope to soon post a blog post and include a link so that I can display some of our monthly data in graphic form which helps for reviewing the data.

The two systems we have installed are as follows:

5K system

Solis 5K inverter S6-GR1P5K

Non-hybrid inverter 

Panels: 11 total; Trina Solar

Max power: 445 watts

TSM-445DEG17M.20(II)

Max Voltage (VMP): 40.8 V

Max Current (IMP): 11.90 Amps 

* No battery system/backup 

System should be grid tied soon. Currently we are generating power such that we are able to use about 40% during the day with the remainder going to the power company so that they can sell it to someone else (never ideal!).

Month of February performance, from start of month to Feb 25:

Average daily yield: 17.28 kWh

Monthly yield: 432 kWh (25 days)

Projected yield for 30 days: 518 kWh 

Daily yield per panel: 1.57 kWh

Peak time for generation: 10 to 11 am

       Approx 4.5 kWh

@12 pm: approx 3.5 kWh

@2 pm: approx  2.6 kWh

2 to 4 pm: drop from 2.6 to 1.5 kWh

Sun is very intense in afternoon but we’re not getting much benefit.

Target should be: 

550 kWh per month or 18.3 kWh per day

My thoughts for improvement to meet target: Add 1 panel. 

Question on efficiency of one panel at 12 pm: 

“One would expect at/near noon on a clear day, the panels should be producing 50% to 100% of their Imp rated current.”

I would say that we are achieving this performance by taking a peak of the power generated around noon:

12 pm

3.5 kWh per system 

3.5 kWh / 11 = 0.318 kWh per panel 

kWh = Amps × Volts × Hours Of Use / 1000

0.318 kWh = y amps x z volts x 0.001

Based on an assumption that z is coming in at 40 volts or a little less, I would say that y is ranging between 8 and 10 amps, putting us in the 60% to 80% range ( back of the envelope without taking an actual reading). 

7 k system 

Solis 7 K inverter 1P7K-4G

Non-hybrid inverter 

Panels: 15 total 

Trina Solar

Max power: 445 watts

TSM-445DEG17M.20(II)

Max Voltage: 40.8 V

Max Current: 11.90 Amps 

No battery system/backup 

System should be grid tied soon.

Average daily yield: 23 kWh

Monthly yield: 573 kWh (25 days)

Projected yield for 30 days: 690 kWh 

Daily yield per panel: 1.533 kWh

  (About the same as other system)

Monthly yield per panel: 46 kWh

Peak time for generation: 10 to 11 am

       Approx 5.2 kWh

@12 pm: approx 3 kWh

@2 pm: approx  2 kWh

2 to 4 pm: drop from 2 to 0.5 kWh

Target should be: 

850 kWh per month or 28.3 kWh per month 

My thoughts for improvement for meeting target: Add 4 panels. 

BB.

OK, the first inverter is a GT Inverter with two MPPT inputs (MP4 connectors)... The basic information being:

https://www.ginlong.com/solarinverter2/2500_6000w_s6_en.html

Input DC for 5 kW model

Recommended max. PV power 7.5 kW 

Max. input voltage 550 V 600 V

Rated voltage 250 V 330 V

Start-up voltage 120 V

MPPT voltage range 90-520 V

Max. input current 14 A / 14 A

Max. short circuit current 22 A / 22 A

MPPT number/Max. input strings number 2/2

https://www.solarreviews.com/manufacturers/trina-solar/solar-panels/trina27872duomaxmtsm445deg17m20ii

Panels: 11 total; Trina Solar

Max power: 445 watts

TSM-445DEG17M.20(II)

Max Voltage (VMP): 40.8 V

Max Current (IMP): 11.90 Amps 

Temperature Coefficient of VOC - 0.25%/°C

You have 2x MPPT inputs limited to 14 Amps each... So a maximum of two strings, one per MPPT channel.
Minimum panels per string: 120 Volts min startup / 40.8 volts Vmp =  2.9 series panels = 3 panels minimum in series
Max panels per string: 600 Volts Max input voltage / 49.4 Volts Voc-std = 12.1 panels in series @ 25C/75F (std condtions)
Derate temperatures to 0C/32F (minimum temperature for your home?): (0C-25C ambient)*- 0.25%/°C = 0.0625 increase in Voc-freezing
1.0635 * 49.4 volts Voc-std = 52.4875 Volts Voc-freezing
600 VDC max inverter solar input / 52.4875 volts freezing = 11.4 panels = 11 panels max in series for moderatly cold areas.
So, for each MPPT channel, you can have 3 panels to 11 panels in series.
Our own suggested "optimum" over paneling = 5,000 Watts / 0.77 panel+controller losses = 6,494 Watt array "nice" over panelling
6,494 Watt over paneling / 445 Watts per panel (std) = 14.6 panels or 14-15 "nice" level of over paneling
7.5 kWatts Pmp-std recommended max / 445 Watts Pmp per panel = 16.8 or possibly 16-17 max panels (for 5kW system)
You have 11 panels, you could have 3 and 8 series panels, 4/7, or 5/6 series strings (5 on one change, MPPT channel, 6 on the other).

Sorry for the "notes style" of math... We can go into details if you wish, but as long as you have 15 or fewer panels, and 3-11 panels per MPPT string, your 5kW system should be pretty close to "optimally" configured. And you presently have 11 panels, you can go to 15 total, or 16-17 panels with some "clipping" (loss of Array wattage around solar noon on cool/clear days)...

The harvest for an average February day with a GT inverter (no battery, no OG/Hybrid inverter), the harvest would be around

11 panels * 445 Watt panels * 0.77 panel+inverter deratings * 5.09 hours of sun (ave February) = 19,185 WH = 19.2 kWH per ave Feb day

All in all, pretty close to your measured 17.28 kWH per day (February?). Pretty much if the calculations and measured output are within 10% -- That is pretty much "dead on identical" for solar without Lab Grade Power equipment (given unknown exact solar conditions, if panels are dusty or not, and typical accuracy of Power/Volts/Current measurements.

Before doing the same thing for your 7 kWatt system (details of panels per string, etc.), the estimated harvest based on array size, Feb Sun, etc.:

15 panels * 445 Watt panels * 0.77 panel+inverter deratings * 5.09 hours of sun Feb = 26,161 WH per ave Feb day = 26.2 kWH per Feb Day

I do have a question is the 17 kWH and 23 kWH per day the "predicted harvest" by your installer, or actual measured harvest? I am a bit unclear as to the "status" of your two systems (estimated output vs operational measured output)...

For some of the numbers above (min/max panels, etc.)--Those are just quick back of the envelope calculations (and should be "conservative"... I have not gone through the manuals for your inverter(s) and I am certainly not trained on your systems. These are just quick "sanity" checks using rules of thumbs and picking some numbers from spec sheets (MPPT range vs minimum starting voltage, etc.) (like you might not get below 20C on a Caribbean island. And you probably do not get too hot either.

-Bill

Keithjwaz

Thanks again for your excellent analysis Bill. Sorry about my delayed response but I was traveling for a few days.

The data that I provided was from the data loggers setup now for our two inverters so it’s live data that has been sent to the cloud. 

Here are some yield numbers for the first few days of March. We’ve been getting very good sun so this Is likely our optimal performance days:

Treehouse 1
5k system

2/28/23. Daily yield: 23 kWh; full load hours: 4.6 h

3/1/23.   Daily yield: 21.3 kWh; full load hours: 4.26 h
3/2/23.   Daily yield: 22.6 kWh; full load hours: 4.5 h

avg yield per day: 22.3 kWh
avg yield per panel: 2.02

Treehouse 2
7k system

2/28/23. Daily yield 27.7 kwh;  full load hours 3.96 h

3/1/23.  Daily yield: 28.8 kWh; full load hours 4.11 h
3/2/23.  Daily yield: 28.9 kWh; full load hours 4.11 h

avg yield per day: 28.46 kWh
avg yield per panel: 1.89 kWh

This shows what we are producing thus time of year under optimal (likely) conditions.

Keith