The greenhouse gas benefits of autonomous vehicles

stanley side view (2005-023-040)

Autonomous vehicles may bring a myriad of benefits, but I anticipate that one of the largest may be the actual reduction in the total size of the vehicle fleet.  Eventually autonomous vehicles will allow “whistlecar” service, and whether fully autonomous or not would, this service is likely to fundamentally change the ownership model of automobiles.  Like present-day car-sharing services or taxis, a whistlecar subscription would mean one car could serve the needs of many people, instead of remaining parked most of the day waiting for its one owner to return.  Once you’re done with a car, it can drive off and serve someone else in the vicinity, drive to a charging station (if it’s electric), drive to a garage for service, or perhaps even deliver packages.  When you can subscribe to an on-demand travel service available 24-7 (and eventually cheaper than owning a car), many people will choose not to own.

Setting aside all the other benefits of autonomous vehicles for the moment, I’ll explore just this one: the benefits of a reduction in the car fleet.  And in a limited way: the greenhouse gas implications of this reduction in vehicles.

According to MIT and the Greenhouse Gases, Regulated Emissions and Energy Use in Transportation (GREET) Model developed by Argonne National Laboratory, the energy and greenhouse gases associated with vehicle manufacture varies by vehicle type, but the MIT report has an average light-duty vehicle figure of 7.7 metric tons of greenhouse gases for a gasoline-powered vehicle.  They have also calculated a 2035 metric, based on some assumptions about materials used to make cars lighter in the future.

The US Energy Information Administration’s Annual Energy Outlook yields loads of projections about vehicle purchases and the vehicle fleet through 2035.  I’ve used these to establish a baseline for the greenhouse gas implications of manufacturing new vehicles.  For simplicity, I assume cars purchased in any given year were also manufactured in that year, which obviously might not be the case, but you get the point.

Here’s the scenario – in 2018, the first whistlecar service roles out in a major US city, perhaps New York, with 1,000 cars.  Uber is an early entrant into the market, again riling the taxi companies.  At first the price is high, regulatory barriers need to be battled, and only some early adopters and “choice” riders adopt.  Over time, with more subscriptions and expansion to other cities, prices come down, the number of vehicles available increases and some people in urban cores with high accessibility give up their cars totally when they can easily access a vehicle on-demand.

I’ve actually developed three scenarios for adoption – low, medium and high, assuming different rates of growth (but all exponential, symbolizing the front half of a technology s-curve).  I also assume whistlecars replace cars on more than a 1-for-1 basis.  This study says car-sharing systems can replace 9 cars per one shared car, so I used that as an estimate.  I ramped up to 9 over time, as people would have to get comfortable with the system, and they wouldn’t be widely available in the beginning. I also assumed that whistlecars would get used a lot more, and would have to be replaced every 5 years.  Finally, I’ve used the 2035 greenhouse gas hybrid-electric greenhouse gas metric for all whistlecars manufactured in any year.


In the high-growth scenario, autonomous on-demand vehicles make up almost all new car sales by 2035.  Only a few thousand personal vehicles are still sold to the very rich, collectors and luddites.  New private car sales have shrunk from a projected 10.6 million 3,500.  Total new car sales (including whistlecars) has shrunk to 1.2 million.  There are close to 4 million whistlecars in the fleet in 2035, replacing 36 million private cars.  This is the full adoption scenario.

In the medium-growth scenario, growth is slower.  There are still 4 million private vehicle sales in the US in 2035 as under the EIA outlook.  There are 2 million whistlecars in the fleet.  Whistlecar sales represent 10% of total new car sales by 2035.  Total new car sales have shrunk to 5.7 million.  This represents a more moderate adoption rate, perhaps some regulatory or technological barriers remain.

In the low-growth scenario, there are only 600,000 whistlecars in the fleet by 2035.  Sales of whistlecars represent only 2% of total new car sales.  Total new car sales have shrunk to 9.2 million.  This represents limited adoption.  Maybe only a few cities in the country have robust autonomous vehicle networks.  Perhaps car manufacturers have lobbied hard to restrict autonomous vehicle use in the face of an existential threat.

So what is the greenhouse gas impact of the reduction in vehicle manufacturing? Potentially quite a lot.

Under the high growth scenario, the reduction in vehicle manufacturing would avoid 77 million metric tons of greenhouse gases in 2035.  77 million mt is more than the combined greenhouse gas output of all the power plants in Minnesota and Wisconsin in 2011.  It’s equivalent to taking 16 million passenger vehicles off the road for a year (that’s 12% of the 2010 light-duty fleet).  It’s equivalent to about 1% of the total U.S. greenhouse gas emissions in 2011.

The medium growth scenario avoids 40 million mt in 2035.  That’s like taking 8.3 million cars off the road for a year.  Low growth avoids 11 million mt in 2035.

If a significant switch to shared, on-demand vehicles is made possible by autonomous vehicles, the reduction in car manufacturing could have a very significant impact on greenhouse gas emissions in the next 20 years.  I haven’t calculated any fuel efficiency savings, of which there could be plenty with autonomous vehicles: lighter weights, less lead-footedness and caravanning.

Some argue that autonomous vehicles will lead to more travel, but I’m not so sure.  By the time there is wide adoption, I think other information technology will offset any increases.  If you can work as effectively in your car as you can in the office, why wouldn’t you stay home?  Plus, how much longer can commutes get?  Robot driver or not, you’re still stuck in a car.  My guess is the above benefits will overwhelm any potential VMT increases by a large margin.

If you want to check my work, here’s the spreadsheet.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s