I recently came across the National Renewable Energy Laboratories Building Typology web dashboard (and associated paper), a very neat interface for exploring a huge simulation of US residential energy use built by NREL (a similar one for commercial buildings is coming later this year.) Using this, we can get a super granular (if simulated) look at exactly where energy goes in US homes, and see which parameters affect it. So let’s take a look and see what we find.
First, a bit about the source. The data comes from ResStock, a simulation designed to model US residential energy use at a much greater level of detail than had been done previously.
Housing in the US varies widely along a huge number of dimensions. Homes vary by size, age, location, materials, type of building, amount of insulation, number of windows, heating system, and dozens of other parameters, many of which are correlated with each other, all of which affect the energy use and performance of a building. This makes it difficult to understand what actually matters for improving energy efficiency. Energy surveys, for instance, break down energy use by building type and by age, but if these factors are correlated (which they are), it makes it hard to know which one matters more. Ideally we’d like to look at age within a building type (and vice versa - look at all building types within a particular age bracket)
To deal with this, ResStock uses a statistical approach. Using a variety of data sources (census data, builders surveys, energy consumption surveys, etc.) they constructed a series of probability distributions for different housing parameters in the US. So, for instance, for housing age it assumes there are X number of houses built before 1940, Y built between 1940 and 1950, and so on. Similar distributions were built for heating fuel used, type of windows, insulation thickness, and many other parameters that affect energy use. These distributions were constructed based on correlations between the various parameters - the distribution of housing ages looks very different in the Northeast than the South, for instance, and the amount of insulation will vary depending on how old your house is. ResStock takes all this into account.
It then samples from this distribution hundreds of thousands of times, and runs each sample (some combination of housing parameters) in a physics-based building simulation to estimate home energy use.
The result is an extremely fine-grained model of how energy is used across US homes. Energy surveys like the Residential Energy Consumption Survey (RECS) tells us how energy use varies from region to region, and across different housing types, but the ResStock data lets us look at different combinations of parameters. How does 1940s housing in the Northeast vary from 2010s housing in the South? How do different states in the same climate region use energy differently? ResStock tells us.
It also gives a more granular look at exactly what energy gets spent on in a home - how much is allocated to heating, to cooling, etc, as well as what type of fuel is used. If you think electrification (replacing fossil fuel appliances and heat sources with electric ones) is an important aspect of dealing with climate change, it's useful to know where that will have the greatest impact. ResStock can tell us that as well.
US residential floor space
Let’s start with a high-level look at US housing. Per the model, as of 2018 the US has about 250 billion square feet of residential floor space, spread across about 100 million individual buildings :
The majority of that square footage is found in single family homes, which make up a bit over 75% of residential square footage. Multifamily buildings (duplexes, apartments, condos, etc.) make up another ~18% of the square footage, and the balance goes to mobile and manufactured homes. Within multifamily, we see the majority of square footage is in smaller buildings - 2 to 4 unit buildings, or buildings 3 stories or less. Multifamily buildings taller than 3 stories make up just over 3% of US housing by floor area - your mental model of “typical apartment building” should be a garden apartment rather than an urban high-rise (the US actually has more square footage in mobile homes than it does in multifamily buildings taller than 3 stories). Over 90% of housing space in the US consists of single family homes and low-rise apartment buildings.
(This all roughly matches with what we found when we previously looked at the US building stock.)
Turning to age, we see a relatively wide spread across different age brackets (which makes sense, considering how slowly homes leave the building stock.) 49% of US residential building floor area was built prior to 1980, and more than 10% was built prior to 1940. The US has more than twice as much housing space in pre-1940 single family homes as it does in large apartment buildings.
US residential energy use
Taken together, those homes use around 11.4 quadrillion BTus of energy annually, which is a bit over 20% (as of 2014) of all the energy used in the US.
So where is all that energy going?
This chart shows the breakdown in energy use (trillions of Btus) and energy intensity (thousands of Btus per square foot) for each building type and age bracket of residential housing. We see that most building types use energy roughly in proportion to their fraction of overall square footage - single family homes, for instance, make up just over 80% of residential energy use, just a bit more than their total fraction of housing space.
We can get a clearer view by looking at energy intensity - how much energy per unit area different types of housing use. We see a fairly uniform decrease in energy intensity with age - no matter the type, older homes use significantly more energy per square foot than newer ones, with the oldest homes using twice as much or more than the newer ones. We also see that mobile homes are the most energy intensive housing type by a fairly wide margin - a mobile home will use somewhere around 40% more energy per square foot than other housing types. (The NREL report makes specific mention of this as something important to address, but I suspect this is mostly a self-solving problem, as these tend to drop out of the building stock much faster than conventional homes do.)
For homes built after 1980, we see that the least energy intensive building type is actually single family homes, something the Residential Energy Consumption Survey confirms. This was surprising to me, since I frequently read stories about how dense urban areas are more energy efficient - I expected multifamily construction to use lower energy per square foot. I suspect the mechanism here is that some types of energy use don’t scale linearly with home size - a 1000 square foot apartment and a 2000 square foot house, for instance, might both use similar amounts of energy for the fridge, dishwasher, microwave, washer/dryer, etc. So smaller housing units use more energy per square foot despite using less energy overall. The 2015 RECS data confirm this - the larger your house, the less energy per square foot it uses:
Overall, average energy use for US homes is about 47,000 Btus per square foot per year. How does this compare to other countries?
Better than you might expect. The average dwelling in France, for instance, uses around 56,000 Btus per square foot of residential space, fairly typical for a European country .
If you instead compare energy per person though, the US looks way worse - we use somewhere in the neighborhood of 38 million Btus per person per year on average, compared to around 23 million in France. Like with single family vs multifamily, differences in cross-country energy use seem to mostly come from differences in housing size, and the US has much larger homes than Europe does:
One confounder here is that European countries often have much older housing stocks than the US, which would skew average energy use higher even if newer homes were significantly more energy efficient than US homes. I couldn’t readily find energy consumption just for recent European construction.
Energy by use
So what is all that energy being used for?
The dashboard breaks down energy use into electricity consumption, and on-site fuel consumption (burning natural gas, propane, or fuel oil.) It then further subdivides it into energy for heating, cooling, water heating, vents/fans, and ‘other’. (This breakdown comes from the fact that the tool was designed to aid in energy retrofits, and so is focused on thermal uses as opposed to, say, energy consumption for kitchen appliances. A more granular breakdown of home electricity use can be found in the base ResStock data.)
Looking just for the moment at single family homes, we see that energy use varies greatly depending on home age: older homes use much more energy for heating than newer homes do:
Pre-1940s housing uses twice the energy per square foot as post-1980 housing. Most of that extra energy comes from heating - ~65% of energy use in pre-1940s homes goes towards heating, compared to ~40% in post-1980s homes, and pre-1940s housing uses more energy per square foot for heating than post-1980 housing uses overall. And the lion’s share of that extra energy use comes in the form of increased on-site fuel burning - natural gas, fuel oil, or propane. Other thermal energy uses make up the second biggest chunk - air conditioning and water heating combined make up another 30% of energy use in post-1980 homes. After thermal uses comes electricity for lights, appliances, computers, etc. (called out as “Electricity: Other” in the graph above), at a bit over 25% of housing energy use in modern housing.
The above is just for single family, but we see a roughly similar energy budget across different housing types. The biggest difference comes from the fact that multifamily housing uses much more electric heating, and less natural gas or fuel oil, than single family homes do.
Energy use per unit area gives us a way to compare different types of housing without the distortion caused by different home sizes. And it’s useful to look at energy breakdown for the country as a whole. But energy use per square foot averaged across all US housing is kind of a weird metric. For one, it lumps in a bunch of radically different housing types together (which is why “average” energy intensity includes both electric and natural gas heating.) For another, there will be selection effects at work - if newer houses were built in milder climates that required less heating, new houses would show lower energy consumption purely by virtue of where they were built.
So let’s drill down and look at housing just in a particular climate region. Here’s energy intensity for single family homes in the “cold/very cold” climate region:
We see the biggest difference is energy for heating - houses in cold climates use twice the energy per square foot for heating than the national average for modern homes, and more than three times the average of pre-1940s homes. Almost all this extra energy use comes from burning fuel on-site. The more granular age breakdown also shows that newer housing continues to get more energy efficient - a house built in the 2010s in a cold climate requires roughly 70% less energy for heating than a pre-1960s house.
How does this compare to multifamily housing?
We see a roughly similar amount of total energy use (though older SFH do worse than older multifamily buildings), but a different breakdown. Multifamily buildings use less natural gas/fuel oil heating and more electric heating, and have lower heating requirements overall (on average, they use about 30% less energy per square foot than single family homes for heating.) But this is offset by higher energy intensity for things like general electricity use.
How does this compare to a less thermally intensive climate zone? Here’s single family use for the hot-humid climate zone (ie: the south):
We see that homes in hot/humid climates use about 70% less energy for heating than homes in cold climates. This is partially offset by increased use of energy for cooling, but on balance hot/humid housing still uses just about 60-70% of the energy per square foot that cold climate housing uses (depending on the year.)
What does it look like if we compare across climate regions? Here’s energy use for homes built post 2010 for all different climate regions in the US. To facilitate comparison, I’ve tried to overlap census regions (comparing different climate zones in the same census region).
We see that energy use varies widely from region to region, and that energy for heating is the main factor that affects this.
Comparing energy codes
Can we get a sense from this data how much energy codes matter? One method would be to compare similar houses built in similar climates in two different states that have different levels of energy code strictness.
The dashboard lets us compare individual states, though it doesn’t let us get quite as granular as we might like (for state-level data, we’re back to pre-1940, 1940-1980, and post-1980 age buckets). We’ll compare Arizona and California, two states that each have portions in the “hot dry” climate zone. California is known as being more stringent with energy codes, having first passed one way back in 1975, and typically leads the way in code strictness. Arizona, on the other hand, adopted energy codes much later (it first adopted one in 2009), and has a history of conservative governors restricting jurisdictions from implementing more stringent ones. We’ll also throw into the mix Mississippi, a state with no energy code. Mississippi doesn’t have any parts in a hot-dry climate, so we’ll look at the data for “mixed humid” and try to scale it accordingly
We see that post 1980 (the most granular we can get at the state level) housing in new mexico is roughly similar in energy use to california. We see that Mississippi uses about 20% more, but this is mostly driven by the increased energy requirements of a different climate zone - control for this and the difference disappears.
But energy code strictness might matter less in a hot/dry climate. What if we look at states in cold climates?
We’ll try comparing New York (long history of residential energy codes) to North Dakota (a ‘home rule’ state that allows local jurisdictions to decide whether to adopt building codes).
Virtually no difference, once again. As far as the simulated data is concerned, there doesn’t seem to be much of an impact on different state energy codes on energy use.
One important caveat here is that it’s very clear that the data in the dashboard is somewhat messy. Different quantities that should equal each other don’t (for instance, if you divide the total energy use by the total square footage for each building type, you get a different energy use per unit area than the dashboard provides), and it seems to miscount quantities in some locations (it appears to double-count single family attached homes, for instance.) In most cases this bust is in the neighborhood of +- 20%, but in some cases it’s much larger - manually calculated values for multifamily energy intensity are half that what the dashboard gives, for instance. I’ve done what I can to cross-check numbers against other data (such as RECS and census data), but we should take any conclusions with a large grain of salt.
So, to sum up:
Modern housing uses substantially less energy than older housing on a per-square-foot basis. A 50-year difference in house age translates to roughly a 50% reduction in energy use. Most of that reduction comes from reduced energy used for heating.
After age, the largest variable that affects home energy use is climate - homes in cold climates use about 30-40% more energy than homes in warm climates, most of which is due to increased heating requirements.
Other than mobile homes, we don’t see a huge amount of variation in energy intensity between different types of housing (single family, multifamily, etc.) On a square-foot basis, single family homes actually use less energy than multifamily housing. Differences in energy use come from the fact that single family homes are much larger than multifamily housing.
Likewise, differences in energy use between the US and European houses stem from the fact that US homes are much larger than European ones.
We don’t seem to find much difference in energy use from across similar homes in states with different level of energy code strictness, though the data here is limited.
The ResStock data in the dashboard is somewhat messy, and care should be taken interpreting it.
These posts will always remain free, but if you find this work valuable, I encourage you to become a paid subscriber. As a paid subscriber, you’ll help support this work and also gain access to a members-only slack channel.
Construction Physics is produced in partnership with the Institute for Progress, a Washington, DC-based think tank. You can learn more about their work by visiting their website.
 - I believe there’s a bug in the dashboard data where single family attached homes are being double-counted, which I have corrected.