# Part 2—Examine Data From a Single Location

Note: This chapter was retired in July 2015 as the tools and data are no longer available. The pages are available here for reference.

## Step 1 – What Does Solar Power Output Look Like?

1. Go to soltrex.com/index.cfm and click the Explore Systems link at the top of the page. Locate the Harvard University system by filtering the list. Enter Harvard University in the search box. Then click on the link to the system. This solar panel installation is located in Petersham, MA at an experimental forest called Harvard Forest, operated by Harvard University scientists.

2. Solar power data is collected in 15 minute intervals from the panels, so there is hourly data, daily data, and monthly data that will have characteristic patterns.
1. Register and/or Login as shown in Part 1.
2. Explore Systems and choose a location and then click on "custom graph" or click the MySoltrex tab, then select "create custom graph".

3. In the "graph controls" orange box, scroll down the system names until you locate "Harvard University - Petersham".
4. Fill in a start date of 1/15/2009 and and end date of 1/15/2009 to look at one day in January.
5. Specify interval "hourly".
6. Choose "Power (kW)" as the data source
7. Click on "Refresh Graph".

Since solar output depends on the sun being out, a solar powered system is not as level a source as other power generation systems where the fuel supply can be more even. If we look at a graph of the power output by a solar panel on a time line of hours across one sunny day, beginning at midnight, what would you expect the graph to look like?

This is a summer day, June 15th, in western Massachusetts and a winter day January 15th at the same location. Notice the difference in peak power. Why might maximum power be less on this June day than on this January day? What else do you notice? [Hint-check sunrise, sunset]

## Step 2 – How Does Solar Power Output Vary by Time Interval?

The next step is to look at different time intervals. Soltrex online graphing has a limit of 1000 data points. There is a point every 15 minutes. This is about 10 days of hourly data or 2.7 years of daily data. Avoid monthly or annual data for now. Look at the online graphs you generate, and then we will make a graph in Excel from a data download in Part 3.
1. Hourly
2. First we will look at power output by the solar panel. This is the hourly aggregate of power over a month in July 2009 in Petersham. Compare this with the solar power output in February right below. Determine whether you think there is more power output in the winter months or the summer months or if it is the same.

This is a northeastern site in the U.S. and at this latitude, the sun is out fewer hours in the winter than in the summer and solar power generation will be less in the winter. You could graph February and July 2009 online at the Soltrex website with interval "month" or download csv files for these two months (interval "day") and sum the daily power to get more exact numbers to compare.

What would you expect measured irradiance data (sunlight) to be at the location of the solar panels?

The sites presented in Part 1 often offer online irradiance data (in units watts/square meter of kWh/m^2), and other weather station data as well. Here is the solar power output (in green) from the solar panels compared to the hourly aggregate of irradiance (in red) over a day, January 15, 2009. It looks pretty similar in shape and start-stop times. Clouds and snow can affect both.

3. Daily
4. Solar panels can be covered by snow and then no power can be generated until the snow melts, which happens pretty readily when the sun comes back out or temperatures rise. Can you see the snowstorms on the Soltrex online graph below of daily solar power for a month in January, 2009?

There were about 4-5 storms. The first low power is actually left from a storm on the 31st of December. The drops in solar power output agree with the weather data for those dates.

what other reasons might there be for a drop in power of a few days?

Strong wind might cause some panel or electronics to break, or disconnect a wire that the logger uses to relay the data. Clouds could reduce the level of power.

Here are daily aggregates of power data over a two-year period. Compare the maximum on the Y-axis in the daily graph below to the monthly data in step 3, below. Why are they different?

The daily values for a month would add up to a value that is plotted for one point on the monthly graph.

5. Monthly
6. A monthly interval on a graph of solar power in kWh over years can show the seasonal difference. In this northern location, you can see the difference in potential power generation between summer and winter.