Tuesday, June 17, 2014

Using data to understand the impact of pollution on PV output

image

Given a recent surge of interest in distributed solar over the past six months, Azure International has received several requests for analysis and data on how air pollution (with PM2.5 as a common proxy) affects PV performance. Two news articles this week have highlighted this important issue and provided varied speculative estimates of the relationship between air pollution and solar output. A short blurb in BJX reports (in Chinese) that when PM2.5 concentrations exceeded 600 in Beijing, PV output fell by 80% versus clear days—but the definition of clear in terms of PM2.5 is not given. A longer article (Cnfol.com, in Chinese) gives a similar 80% output drop for a system operating in Shanghai between two similar days in December 2013, one with clear weather and one with PM2.5 concentrations exceeding 600. The latter article also uses NASA data to compare solar radiation in Beijing in 2012 versus 2013, which was more polluted; solar radiation fell 10% between the two years.

There are many factors that can potentially affect the relationship between air pollution and PV performance. Cloud cover, moisture content, ambient air temperature and other climate conditions make such comparisons difficult. Nor is PM2.5 the only type of air pollution that would affect PV output. Cleaning schedules also matter. Making the analysis even more difficult is the difficulty in finding reliable PV performance data, especially since most distributed solar in China has such a short track record. Below, Azure utilizes a model we have developed in house to evaluate  solar attractiveness by province, using PM2.5 levels as one of ten evaluation criteria.

China urban PM2.5 levels by province

image

Source: Azure International, based on 2013 city PM2.5 averages by province (MEP data)

Lohaus in Shanghai has been tracking the power output of a 1 kW rooftop system since early April 2014. Azure ran correlations on the daily output compared to PM2.5, cloud cover and humidity, but so far data are insufficient to draw firm conclusions. However, the system recorded fairly good output on several clear, low-humidity days when PM2.5 concentrations  were in the range between 100 and 200.

In addition, Greenpeace has supplied Azure with data on its 5 kW Shunyi warehouse rooftop PV system’s daily performance in December 2013 and January 2014. Fortunately, those two months showed very little cloud cover during daylight hours, consistent temperatures averaging near freezing, and fairly low relative humidity—reflecting Beijing’s typically clear, dry winters. As with the Shanghai data, there are insufficient data to clarify the mathematical relationships between temperature, relative humidity, cloud cover, PM2.5 and PV output. However, Greenpeace’s own analysis  shows a clear drop in output due to PM2.5, and on one day of high pollution this year, the system produced no output at all, worse than even on a typical cloudy day. Azure’s analysis shows that on days with PM2.5 concentrations above 100, the system’s average output was half that of days when the PM2.5 concentration was under 50. Concentrations between 50 and 100 showed an average reduction of 20% versus concentrations above 100. (Stay tuned for Azure’s full blog update on our visit to Greenpeace and quantitative analysis of Greenpeace data.)

Greenpeace solar system output (kWh) under different PM2.5 Concentrations

image

Source: Azure International, using data from Greenpeace, U.S. State Department, and Weather Underground

Anecdotally, several sources have reported to Azure that severe air pollution reduces PV output by about 15% on an annual basis. Average annual PM2.5 concentrations in Hebei’s major cities are around 100. Scaled linearly, and without controlling for any other factor, this implies a 14% reduction in PV output on an annual basis. Azure’s earlier analysis of Beijing PM2.5 hourly data looked at PM2.5 patterns by time of day and season of the year, fitting a regression curve for these factors that proved highly significant. Happily for solar, summer and mid-afternoon are relatively less polluted compared to other times, though the biggest difference is actually in winter.

Beijing historical PM2.5 concentration by season and sunlight hours (8 am-4 pm)

image

Source: Azure International

Taking only these factors into consideration, a simplified adjusted solar radiation curve by month would suggest a 12% impact of PM2.5 on an annual basis. This estimate is little more than a best guess approximation. Analysis of actual solar output over several years in a given location would be needed to form a rigorous estimate.

Bejing PM2.5 concentration and horizontal solar radiation by month

image

Source: Azure International

iSource for top photo of Greenpeace solar system in Shunyi, taken by Julian Schwabe, China Greentech, June 16, 2014, and used with permission