This section covers the Sunlight Estimate used in PV System Design. It is background information showing how the number was obtained, it may be more detail than you care about. The Sunlight Estimate I used was 4.4 hours per panel per day. See Generating Power for a revised Sunlight Estimate based on actual data from my PV system after it was installed.
PV panels are specified as generating a certain amount of power (e.g. 200 watts) at a certain light level. That light level is, by convention, 1000 W/m² (watts per meter squared), which is about the light level due to the sun at the earth’s equator when the sun is directly overhead. (So a 200W, 1m² panel would be 20% efficient)
However Kona is not at the equator, and the sun is not directly overhead all day (in fact it is only overhead for a moment at ‘Lahaina Noon’). So I need a ‘Sunlight Estimate’ for my location in order to estimate how much power a PV panel will actually generate.
Determining the Sunlight Estimate
The actual number 4.4 h/panel/day comes from using PVWatts (I found using this tool far from intuitive, but I followed the instructions and the result is worth the effort).
PVWatts will report incident solar radiation for anywhere in the US, it assumes a de-rate factor for PV systems which takes into account all the energy losses that will occur between the surface of the panel and the AC disconnect switch (the default value it uses is 0.77 and its constituent parts are listed), and a retail cost of electricity (which I had to change to reflect current prices). I also specified a 2kW of panels representing my system.
When running PVWatts I selected South Kona as a location, and the tool reported (see table below) that the incident radiation is on average 5.79kWh/m²/day. Multiply by the de-rate factor of 0.77 and the panel area of 1m² (actually 1.18 in my case) and I got 4.46h/panel/day, which I rounded down to 4.4h/panel/day. What I hope is a conservative but reasonable number to use in system design.
This 4.4h/panel/day number is for a panel with 1m² ‘active silicon area’, I know the physical area of my Sanyo panels is 1.18m² and that this is not all active silicon. I am assuming 85% of the area of the panel is silicon, looking at the photograph and image from the data sheet below, this may be an under estimate; resulting in a Sunlight Estimate that is under estimated. Under estimating the sunlight means that in the actual usage my panels will produce more energy than I expect, this will make me happy during actual operation, so the underestimate seems like a good thing!
 |
 |
This was the information PVWatts reported (I have highlighted the significant parts in red):
|
||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
This data is compared to actual inverter data from my house in Generating Power. That section contains a revises Sunlight Estimate based on actual results.
The map PVWatts displays suggests that it does not know about microclimates, so this may not be exactly the conditions at my house. If fact I know this is the case because the table says August is the month with the most solar power, given that August is rainy season here I don’t think this is accurate. But I do believe it is the best basis around for system design, but probably not exactly correct. PVWatts also tells me that 2kW of panels generates $894 worth of electricity a year.
As an alternative design approach it might be interesting to look at the solar maps that the solar hot water installers use. They are, I think, calibrated in bizarre units that plumbers understand like calories/inch²/week (or whatever) which would have to be converted to kWh/m²/day; I did not investigate this approach. I notice that despite having fine grain solar radiation maps, solar hot water guys put two 80ft² panels at almost every installation – perhaps size really doesn’t matter that much!
A ‘How To’ for Grid-Tie PV on the Big Island of Hawaii
copyright © 2006 - 2009 Robert Flatt, website by jasmarlin.com
