* 2/3 of the panels were offline for maintenance and replacement from March to July of 2008. 2008 is not included in the average and standard deviation for AC generated.|
Data for 2011 and 2012 has been requested from the SFPUC.
For 2009, insolation was lower than the average of the previous five years. Output and performance ratio were the highest ever and were finally better than predicted.
Here is a more detailed look at Moscone Center insolation.
- Possible financial bottom line: cost of system, $4.2 million before rebates, $6,222 per kW system. Average output 2004-2007, 1,126 kWh/yr per kW system x $0.12/kWh = $135 of electricity generated per year per kW system, equals 2.2% return on investment.
- Predicted output (2004) was "at least" 826,000 kWh per year, which is higher than the results obtained. However, the predicted output was based on the NREL (National Renewable Energy Laboratory) 1961-1990 figure for solar insolation for San Francisco of 4.72 kWh/m2/day. The 1991-2005 NREL data figure is 4.46 kWh/m2/day, which is very close to the average insolation at the Moscone Center for 2004-2007. As shown in the chart above, the Moscone output corresponds to 4.49 kWh/m2/day input and a more conservative 69% performance ratio, rather than the predicted 71% performance ratio.
- National Solar Radiation Database 1991-2005 Update: User's Manual (PDF, 14 MB) page 2: "Nearly all of the solar data in the original NSRDB and this update were produced using solar radiation models. Less than 1% of the records in this update contain measured data." The decline in predicted insolation at San Francisco from 4.72 to 4.46 kWh/m2/day is statistically significant. It may be due to an error in the 1961-1990 model.
- If the SFPUC online solar savings calculator is used with the inputs Incoming Solar Energy (GHI) 1,639 kWh/m2/yr and System Size 675 kW Peak, the predicted result is: Annual Output 743,782 AC kWh.
- The panels at the Moscone Center lie flat on the roof, therefore sunlight never hits the panels at a right angle that would be optimal for power generation. The closely spaced panels are difficult to keep clean.
- Inverters draw power in their standby mode and the Moscone system has three inverters, so negative outputs would be expected at night. However, the Moscone system does not have bidirectional meters, so no negative numbers appear in the SFPUC data. For each of the three Xantrex PV225 Grid Tie Inverters installed, the Standby Tare Losses are 107 watts, so readings of approximately (-)300 watt hours would be expected every hour at night or other times when there is no sunlight. As there is no sunlight for approximately one-half of the hours in a year, this could be a loss of more than 1,000 kWh per year. There may be other power losses that are not recorded.
The fine print: Historical data is obtained from spreadsheets kindly emailed to me by the SFPUC.
The performance ratio is output/input,
(AC kWh generated per DC kW system)/(insolation kWh/m2).  This figure does not include the 12.1% conversion efficiency of the solar tiles used at the Moscone Center.  Conversion efficiency is (DC kW generated)/(m2).  System efficiency is (total AC kWh generated)/((insolation kWh/m2)* (total m2 of panel)), and is 8.6% to 9.1% for the above data. System efficiency may also be calculated as (conversion efficiency) x (performance ratio).