Sizing solar panels: 1600W consumption and battery bank recharge

One derating factor that you may have missed... 80% flooded cell battery efficiency or 90%+ AGM battery efficiency.

There is also one other derating factor that you could plug in--That is the ~95% charge controller efficiency. You used NOCT which is ~72% for that panel--I usually use 77% derating for panels+controller--So you are already a bit more "conservative" that I normally use. However, if your system is in a very hot climate and installed near sea level/humid area--That 72% derating (NOCT) + another 95% derating (controller) may be needed.

• 33 panels * 1/0.90 AGM battery derating * 1/0.95 MPPT charge controller derating = 39 panels

And if you do the same 0.95 derating for your panels to run the load, then:

• 8 panels * 1/0.95 controller derate = 8.4 panels -- Not much different

Otherwise, your calculations look OK. If your loads are one for 1-2 days, then off for a day or two--That does give you battery bank a chance to fully recharge--There are just not enough hours in a day for a Lead Acid bank to fully recharge from 70% DoD every day. Of course, ideally in sunny weather, you are only running from the battery bank ~18 hours per day and from the solar array the other ~6+ hours.

I still believe the 6 hours of sun per day for a fixed array is a bit optimistic for most regions of the world except for high deserts. You cold add tracking, or more panels, or split your array with 1/2 the array tiled east and the other 1/2 tilted west (virtual tracking--You would need a somewhat larger array for same output and "more charging hours per day", but (possibly) cheaper than mounting your 1/3rd fewer panels on full two axis array--) or you could go with "real" 2-axis tracking. Just to give you an idea of how much that could affect your output, using PV Watts for KITALE, KENYA, flat array vs a 2-axis tracking:
http://pvwatts.nrel.gov/pvwatts.php

Flat array (zero degree tilt)

Month	Solar Radiation ( kWh / m2 / day )
January	5.81
February	6.19
March	5.77
April	5.16
May	5.18
June	4.89
July	4.87
August	5.20
September	5.63
October	5.55
November	5.26
December	5.57
Annual	5.42

2-axis tracking:

Month	Solar Radiation ( kWh / m2 / day )
January	8.09
February	8.52
March	7.30
April	6.31
May	6.64
June	6.38
July	6.08
August	6.51
September	7.10
October	6.95
November	6.75
December	7.68
Annual	7.03

The above is based on "real data" for an equatorial system. Could do some more work with a spread sheet of the hourly data, split east/west array, and see how many hours of "useful" sun you can get.

At this point, solar panel are the "cheap part" of your system (vs battery bank). I would try to error on the "larger array" side. That will give you more flexibility and a "happier" battery bank. Will also give you some extra capacity in case of failures (failed panel, battery cell/bank capacity issues, scientists want "more power").

-Bill

If you can swing LiFePO4 batteries--That would be great. Are you looking at a Battery Management System for the LFPO batteries? Charge controller needs to have zero temperature compensation and programmed for max voltage. You also need min-voltage power disconnect/shutdown (or similar) to prevent ruining your bank.

Will the battery/power system be in a separate building? Or at least a sealed off room from the rest of the equipment. Battery (and genset) fires are not common, but reducing your exposure to fire/smoke for expensive equipment is not a bad thing.

Can you gives us a Longitude as well? That can tell us a whole bunch more about the weather altitude/etc.

If this is a long way off (year or more)--Can you get a weather station + solar energy for logging out there? Or even a USB logger and a solar cell would tell you a lot

http://www.davisnet.com
https://www.dataq.com/resources/pdfs/datasheets/el-data-logger.pdf
https://www.lascarelectronics.com/easylog-data-logger-el-usb-4/

$80+small solar panel (a "shorted solar panel", i.e., measuring current will give you solar radiation within ~5%-10% accuracy or so).. 1 year battery life, 22 months of logging with 30 minute sample.

Whole thing the size of your hand (or smaller).

-Bill

You could even build a small scale model of your system and install it out there--And log/monitor the results. A nice/small MPPT charge controller and a couple hundred watts of panels--Plus some sort of programmable load on the battery to simulate your proposed profile.

Morningstar SunSaver 15 Amp MPPT Solar Charge Controller ($245)

-BB

You will want your loads to shut down. Not the battery bank (would require manual intervention to recharge). Over and under voltage can immediately ruin LFPO cells.

Our host, NAWS, (and other companies) can provide design support and various levels of preconfigured systems.

Also, there are "epanels" that are prewired for various charge controllers and inverters.

You will also find a number of discussions on sailing and RV forums about LiFePO4 batteries too.

- Bill