Data - dssg/energywise GitHub Wiki

Our energy analytics tool consumes building energy data and analyzes it to understand a building's energy behavior.

To develop this tool, then, we needed real building data. There are two ways to get it: utility companies have interval data for the buildings to they supply power to. If you are a direct client, you can access this information with your account number. Or you can purchase anonymized building energy data from the utilities.

Agentis energy supplied us with anonymized interval data they purchased from a utility company.

Building interval data from Agentis

Agentis energy is a company that provides utilities with a software platform for their business customers. They use analytics and software to improve costumer satisfaction and efficiency in commercial buildings.

With their kind support, we received hourly meter data (electricity usage in kWh) for roughly 7,000 buildings. Agentis also gave us the NAICS code and business type for each building. We used this data to develop our tool. Because the data was anonymized, and we didn't know the location of these buildings, Agentis provided us with the associated temperatures for the same time series.

The NAICS code proved to be a very important feature in our analysis because, in using it, we were able to create measures of comparability across buildings based on the raw data relying on supervised methods.

Outside temperature information is very important because it affects dramatically energy consumption when people use the building: it either needs to be heated or cooled.

The raw interval data for one year is shown in the following figure.

The blue line represents hourly energy usage in kWhs for a single building. Sometimes there is no information for a particular hour, which we depict in red. Every point in red was linearly interpolated (although, the statistics calculated do not use inputted values).

Depending on the utility that provides the information, the data usually comes in a csv, excel or xml format. To work with the data, we created a python script that transforms this data into a dictionary. The advantage of working in this way (over a traditional database) is its flexibility to accept building variables as they become available.

So each building record is a dictionary that contains keys for building type (single value), NAICS code (single value), temperatures, kWhs and times, and more fields could be added. The times is a numpy array with datetime objects. The temperatures and the kWhs are masked numpy arrays with the value corresponding to each of the times (we mask them to know when we had the value and when we inputted it). If a developer has further information relevant to the study of a buildings' energy consumption, the building record can easily be expanded to include it.

Building interval data from the State of Illinois Prisons

Once our tool was built, we used it in a case study with additional building energy use data. The State of Illinois is looking for ways to improve their efficiency in energy usage. So we worked with the Department of Corrections, who provided us with interval data for several of Illinois' prisons. We used Energywise to investigate the energy savings potential of these buildings.

These facilities proved to be ideal for a case study for two reasons. First, each facility is very similar in footprint and purpose, making it easy to compare apples to apples. Second, as correctional facilities, engineers have nearly total control over the electrical systems (as opposed to office buildings, where individual employees may set individual schedules and temperature preferences).

The report performed on this case study is to be presented to Illinois Correctional Facilities. We would like to thank them for their data and for their interest in becoming more energy efficient.