Blog Archives

5 inches of rain. Big storm, strong, gusty winds. Wet, wet, wet. That is the prediction over the coming weekend.

I have been meaning to write a follow on to my hooked on solar entry, but have been stymied by the writer’s block, most likely brought on by Shanthala’s return to a punishing, unrelenting schedule. I wanted to put up an entry that showed how the solar panel performed over the course of a year. How close was the actual production to the specifications ? What factors affected the performance ? How did the performance vary through the day ? Through the year ?

A quick recap of the story. In the first year after the installation of the panels, our electricity bill for the entire year was $4.68, not including the monthly minimums (about $2 or so per month). The solar panels are 13 panels of Evergreen’s ES-A 205 panels with a Fronius IG 3000 Inverter. The solar panels were installed by Solarcity. As part of their service, they provide monitoring service with all sorts of relevant information captured and made accessible over the web. The information is captured by a device and sent every 15 minutes to the monitoring location. Via a web portal, I can access either graphs that summarize the relevant information or get the raw collected data. The following charts were taken from the web portal back in November, when I planned the followup to my original article on installing solar panels, so they’re a little dated, but since they tell the story across a year, their current age shouldn’t alter the story by much.

The Year’s Story

Here is a graph that shows the production over the past year. The peak production for the year occurred, naturally, in the summer months of June (490 kWh), July (496 kWh) and August (465 kWh). The worst production has been in the winter months of January (177 kWh) and February (225 kWh). So, the best months produced about double the energy of the worst months.

A look at the month’s data (the best and the worst) reveals:

During January, there is barely 10 hours of daylight and the cloud cover is an average of 51% (which hides the data that had some days of 100% cloud cover and a couple of days of sunshine) while in July, daylight hours average about 14.5 hours with an average cloud cover of only 23%. Only cloud cover during the daylight hours are taken into account in this statistic. So, you have days in January that barely show up in the chart while most days in July have energy production consistently upwards of 15 kWh.

Seasonal Variance

However, if you look at two cloudy days, with the same amount of cloudiness (85%), one in the pitch of autumn and one in almost summer, the production is still quite significantly different as the following two charts show. The hours of sunlight and that the sun is higher in the sky during early summer probably explain the entire difference.

Other similarly cloudy days in May got far better performance as this graph shows:

If you compare a single bright sunny day’s energy production during a summer and a winter month, you get the graph below.

Sunrise was at 7.22 am on the winter day and sunset was at 5.02 pm. The corresponding numbers for the summer day were 5.55 am and 8.14 pm. Energy production peaked around noon on both days. Both the rise and fall in production are quite precipitous. The yellow line shows the production on a day when the cloud cover was 100%. The line barely manages to lift itself off the ground.

So, the lesson from this is that even though the total panel rating is 2.665 kWh, this number is produced around a very narrow spectrum of hours in a day (about 5.5 sun hours, according to this document). Further, the conversion from DC to AC (this is measured at the inverter) cuts the peak power from 2.665 to around 2.25 kWh, about a loss of 15%. According to this document on solar panel performance produced by USREA, system wiring and inverter losses by about 11%. System wiring losses include reduction due to varying performance of the individual panels. If a panel is rated as being 205W +/- 5W, then the production of that panel can be 200W or 210W. Eventually, the total production drops to that of the lowest performing panel according to this document. Evergreen panels state this problem specifically and say that is why they endeavor to keep their panel performance with only a possible upside, never a downside to the production i.e. our panels can produce 205-210 W per panel, but never lower than the rated 205W.

Actual Vs Predicted Performance

There are various tools out on the web that provide some measure of predicted performance based on some installation specific information and solar radiation data gathered over 30 years (1960-90) by NOAA. One such site is US government’s NREL(National Renewable Energy Lab)’s site on solar with its calculator called PVWatts. The detailed installation information provided by SolarCity enabled me to enter all the requested information to obtain the predicted performance data. Here is a chart that shows the predicted vs actual production (not including December) using San Francisco’s data (the closest point to Sunnyvale in the charts):

Like everything else that uses past data, the tool warns that it uses averages of the data and that the performance of any particular year can vary from the predicted performance by as much as 30% (+/-) on a monthly basis and +/-10% on a yearly basis. Still, like an investor looking at a single year’s return of his funds feels thrilled that they have performed well compared to the market indices, I felt somewhat gratified that the panel’s performance over the year has been a little better than the predicted performance. However, Sunnyvale is sunnier than San Francisco with a lot less fog. Oh! Well. The tool provides equivalent data if you stay in some other part of the country, say the Northeast or the Midwest.

Other interesting data provided by the SolarCity include the amount of CO2 offset by the solar panels (I presume this is based on assuming the saving had the equivalent energy been produced by conventional fossil fuels by the electric company) and the dollar amount of energy produced since the time of installation. As of today, these data are that the panels have offset 5,805 lbs of CO2 since installation and produced $816 worth of electricity. Incidentally, 5,805 lbs of CO2 is the equivalent of what 2.8 mature trees would offset over the same duration. Wow! All the CO2 that is offset by all this money (the cost of installing the panels) is matched by just 3 mature trees ?

References:

• Solar Module Performance
• Solar Detective: One of the postings from Scientific American’s Solar At Home blog that investigates the reasons behind a panel’s production
• NREL’s Solar Resources Page
• Going Solar: One Year Later: Another story by a panel owner in Sunnyvale