Categories

Tags

← Blog Home

50 Years of Consumption and Efficiency - An Exploration of the TrueHome Simulation Model

From social media to energy efficiency, data science is a widely played sport. After years on Twitter's data science team, I recently came to Tendril to create interactive data visualizations with a focus on stories that explore energy analytics. In an effort to familiarize myself with our platform, I leveraged some of the data points from Tendril's TrueHome Simulation Model to create my inaugural data visualization. In creating this visualization, I set out to better understand our model’s performance and to see how a sample set of significant home features varied across different weather regions. I selected a 50 year period as I felt this might show the most drastic fluctuations in energy consumption and home efficiency. You can use the interactive pieces throughout this blog.

An Introduction to Our Model
Starting with a nationwide study accomplished with NREL, Tendril built a basis for understanding building materials by home location and vintage. This study provides what we call our model’s “smart defaults”—settings that can generalize the features of homes in a particular region. Keep in mind, home design has changed over the years and accordingly, so do the model’s default settings for each year. For instance, the insulation quality of a home built in 1950 is likely to be very different than the insulation installed in 2000. It’s worth mentioning that these defaults can be overwritten when additional information becomes available about a specific home.

Additionally, all of the U.S. addresses entered into this model are associated with one of the many weather stations around the nation. These stations provide detailed information for the climate in a specific home location. The model uses these weather patterns to consider demand for energy sources. For example, a home in Florida would predictably use more electricity in the summer due to HVAC cooling.

Visualizing the Data
First, I used voronoi partitions to approximate weather regions around U.S. weather station locations. I then colored the map with a gradient to represent consumption levels in each weather region. This map is showing the range of electricity consumption for each weather region based on the year in which the homes are built. You can change the vintage and hover on the weather regions to see how the consumption levels vary.

To further understand the inner workings of our model at a regional level, I selected a sample set of each weather region’s default home features. I then created a single line curve for each region. The regional curves demonstrate each feature's efficiency ratings. The line curves are colored-coded by each region's corresponding consumption level from the map. Note that these regional curves overlap when the select default settings are the same. The further the curve is to the right on each axis, the more efficient the home can perform in our model (across these select home features). In doing this, we can see how regional defaults compare to one another.

Finally, to understand the efficiency scores at the national level, I added bar graphs to each efficiency scale to see how many regions are grouped into a specific efficiency score. A large bar to the right of the scale would indicate that most regions in a particular vintage contained a more efficient feature rating.    

Physics-based vs Regression Modeling
It’s important to note that since our TrueHome Simulation Model is physics-based, home features do not simply shape the model in a finite way, as they would with a regression model. Regression models apply static parameters to home features without taking into account the other features within the home and how they may affect one another. In contrast, our physics-based TrueHome Simulation Model shows the home as it actually performs in the real world. Different home features work in conjunction with one another and can even affect each other, so our model is built to represent those interactions. For instance, the number and type of lights will have an impact on the air conditioning load of a home. As we look at the colors on the map representing energy consumption along with the dancing line curves representing efficiency that varies through the years, keep in mind the complexity of a single feature impacting the overall consumption level.

Observations
As I combed through the map and line curves, a few observations emerged. We see higher electricity use in the southern US, which makes sense at that latitude since we can infer it is largely due to air conditioning. Once we switch the map to show gas/oil, notice how the consumption in the northwest near Washington and Oregon is generally much lower than the northeast near the same latitude in Vermont and Maine; the opposite is true in the 1950s and 1970s if we toggle to electricity. I expected to see these four states using mostly gas or oil to heat, without using much electricity. So why do Oregon and Washington have significantly higher electricity consumption in those two periods? Homes in the northwest area were in fact built with electric heating—a finding that was not immediately apparent without our visual story.

As we cycle through the years, homes generally increase in square footage, which is why we see a decrease in their size efficiency. Interesting to note is that although homes are getting bigger, the remainder of the default settings indicate that they are moving further to the right and generally improving their energy efficiency. For example, notice that the insulation ratings (measured via r-values) increase over time in the home features (roof, windows, walls, floors), which means that the insulation quality is improving.

My first visualization was a successful exploration of our model’s performance and allowed us to see how a particular set of significant home features varied across weather regions for differing home vintages. I was even able to decipher why regions on similar latitudes would have vastly different electric vs oil/gas consumption levels. But this is only the beginning. Stay tuned as we dive deeper into the TrueHome Simulation Model to create and share more data driven visualizations.

You can experience the full interactive visualization below. 

Related

Categories

Tags

  • Continuous Demand Management
  • Customer Ops & CSAT
  • DERs
  • DSM
  • Data Analytics
  • Demand Response
  • Disruption
  • Energy Efficiency
  • HERs
  • High Bill Alerts
  • Privacy & Security
  • Smart Home
  • Solar

    Invalid Email Address

    Subscribe to This Feed

    Thank you for subscribing to Behind the Meter Blog.