Pholtovoltaic RoofingWe have chosen to use SolarSlates from AtlantisEnergy. The slates are rated as 13.3 watts/slate or 10 watts per squarefoot. Their revised page gives specifications for their slates and give an idea of what is available. Their installation manual was quite helpful. (Their webpage has been revised and gives some very nice examples of their product.)
We have made sure that Atlantis understands that we are doing grid interconnect with battery backup. The issue of how many slates to connect in series is related to what kind of output we are using. We have installed sets of 24 slates/field.
InstallationWith wonderful cooperation from Atlantis Energy and a grant from the Illinois Department of Commerce and Community Affairs we took delivery of 480 solar slates in November, 2002, and installed them in early December on three different roof locations: The garage, with a pitch of 18 degrees and the Belvedere, pitched at 33 degrees, both are directly south facing. The 2nd floor roof, pitched at 30 degrees, faces SSE (150 degrees). The number of slates are 216 (garage), 120 (2nd floor), and 144 (Belvedere). Installation went very well with the help of Atlantis Energy's installer instructor. (For more than enough photos of the installation on the garage and Belvedere, see the photos page.)
We now have been hooked up to the Commonwealth Edison grid since January 28th and thus have been able to measure output for more than 6 months. As of September 10 total output since starting up (measured by the C40 charge controller) has been 4,325 KWH or slightly more than 19KWH/day. This leads to a predicted annual production of almost 7,000 KWH which is almost 30% less than I had expected given the 1.5 KW/W prediction. This could have been due to problems with the inverters (if they are not inverting, the batteries fill up and the charge controllers shunt any power from the arrays) or it could be a problem with the installation of the slates, or the slates themselves. Clearly more investigation is necessary.
Our production can also be expressed in pounds of CO2 that we have saved. Assuming .7 lbs of CO2/KWH -- the Illinois value is lower than the 1.3 lbs in other states because Commonwealth Edison uses nuclear fired generators -- and 19.5lbs/gallon of gas, we have reduced our carbon load by 1.5 tons or the equivalent of 3,000 miles of driving at 20 mpg.
I have been recording data manually from the C40 charge controllers about once per day and the slates are working very well. Total daily output has ranged from 4.5 Kwhr (on a snowy day) to 32 Kwhr (on a very clear day). (Note that occasionally one of the inverters seems to need to be manually reset and until we do so, we are not selling power. This led to a very low reading for one day in May and several days this summer. There seems to be a tendency for one of the inverters to fail if running on batteries for more than 10 minutes. We are investigating this problem, and it seems to have been fixed.) We have three different roof fields, at different pitches, orientations, and sizes. The maximum output per field ranges from 7.9 (2nd floor) to 14.8 (garage) Kwhr/day. Note the variation as a function of weather and the long term trend as spring arrived. Snow obviously inhibits production. This effect varies by roof location as the slopes collect and shed snow at different rates. But the fields differ in the number of slates and so a better measure is the production per slate which also varies as a function of location as well as day (snow load). The highest output has been from the belvedere slates which have produced from 8 to 72 watt hours/day/slate. The garage slates, with a lower pitch, have produced from 3 to 68 watt hours/day/slate. I have occasionally observed the instantaneous output of the three fields, and the highest output per slate has been 12.0 watts. Output varies throughout the day, with substantial output between 9am and 4 pm. On a weekly basis, we have been averaging more than 20 kwh/day for the past several months, as would be predicted from the greater elevations of sun as winter has passed.Now that we have more than 6 months of data, it is useful to observe monthly averages of the total daily output, output per slate/field, and to compare consumption to production. The seasonal trends are very apparent in these charts, showing a peaking mid June at the summer solstice (just as we moved in), and a gradual falling off now that we are in September. (Comparing daily, weekly and monthly data graphically is a nice demonstration of the power of smoothing data to see trends. It is also interesting to observe how much consumption has fallen off once very active construction stopped.)
Also available from the records is the observation that our gross consumption of electricity is averaging about 16.2 kwh/day, which, with a gross production of around 19 kwh, is leading to a net sale to the grid of around 3kwh/day. We have brought our average hourly consumption (averaged over the week) down to about 680 watts/hour.
Further data on electrical operationUnfortunately, the above analyses were written at the end of last summer when I was falsely attributing the high consumption in winter to being under construction. With another winter behind us, I now recognize that our power demands during winter are substantially higher (roughly 60-100 watts/ degree day) and that our production is much lower than expected. We had a very cloudy winter and early spring, more so than normal, so perhaps this is an undue pessimism. I have added a more detailed analysis of our power production and demand and will be updating this page when I have time.
Electrical EquipmentBased upon the recommendations of Vladimir Nekola of Nekolux consulting, we determined to use the following equipment. Most of it is provided by either TraceEngineering (of British Columbia) or Outback Power (of Washington). It turns out that some of these recommendations were not as well researched as we would have hoped, for the battery cabinet recommended (PSR) is not well sized for the batteries we are using, nor were we told that we needed a grid intertie interface. It also turns out that at least one charge controller was undersized for the expected output. Difficulties such as this are probably not unexpected when installing products that are not quite "off the shelf". Part of the purpose of this webpage is to help others by sharing what we have learned by out successes and mistakes. We are in the process of replacing on C40 charge controller with a MX60 from Outback.
Expected output of 8G8D JellCell batteries is about 1140 Watts/hour for 6 hours or 388 Watts/hour for 20 hours.
According to the Trace web page, we also need a Grid Tie Interface to interact with Commonwealth Edison.
Trace also offers a GridTie system (not involving batteries). There is a nice introduction and review of how it works. This include a discussion comparing pure grid tie with grid tie with battery backups.
with the following graphics taken from their page (but which have since disappeared).
The batteries (8G 8D JellCell) although marketed by Schott are actually produced by East Penn Manufacturing. The East Penn web page discusses the need for some venting of the batteries, but this seems to mean just venting to the room. The batteries should not be in a sealed case but rather in a case that has some venting.
In addition, the batteries are very sensitive to the voltage and temperature at charging. The pdf file from East Penn (go to this page, select the pdf file to download, fill out the registration form, and then you have the pdf file) provides exact temperature and voltage requirements.
Note that the charge controller provides a thermometer to monitor the battery while charging.
Note that because these gel cell batteries are lead based, they need to be recycled, a service provided by East Penn when more batteries are purchased.