I spent a few of the Sandy storm days indoors and decided to indulge my inner nerd. I’ve been tracking all our house performance data using monthly spreadsheets, which can make it difficult to filter data or look for trends across months. So I downloaded mySQL and imported all our energy and temperature data into the database. This is the first in a series of posts on what I find as I sift through the data.
A while back I stumbled upon this article, Linear Regression Analysis of Energy Consumption Data. At the time I was learning about how to calculate HDD days based on our own temperature data captured by our outdoor HOBO temperature and humidity sensor. Now that our data is much easier to query I decided to give the linear regression analysis a try.
The main idea is that if we plot energy used to heat our house on one axis and HDD on the other (and we’re extremely lucky) we should be able to see a fairly consistent pattern. Meaning we could use the results of the analysis to predict how much energy will be required in the future if we know the probability of HDD for a future time period.
The R2 value of 0.1134 in the chart at the top of the page tells us that we don’t have a strong correlation between energy use and HDD. The closer the R2 value gets to 1, the stronger the correlation. So what could be the reasons that we don’t have a stronger correlation?
One reason may be that an air source heat pump uses different amounts of energy to make heat depending on the outside air temperature. It has to work harder at lower temperatures to generate the same amount of heat.
It may also have something to do with how the HDD values are selected. I started with daily snapshots, but there could be 23 HDD in a 24 hour day, and the ASHP only operated for a few of those hours, either because we turned it off at night or we experienced a day with a lot of sun. So I switched to hourly snapshots and discarded any hours where the ASHP was not drawing power (the result is the chart at top of page). This however, did not result in a closer correlation.
My guess is that we don’t use the heat in a consistent manner so we’d have a difficult time getting a strong correlation. Plus, the way the heat pump itself works may also make it difficult to correlate.