When Should You Open Your Car Windows? An Experiment

Back at the start of the summer, I asked a question about automotive thermodynamics: On a hot day, is it better to open your car windows a crack when making a short stop, or leave them closed? For a long term-- say, leaving your car parked outside all day-- I hope everyone will agree that leaving the windows slightly open is the better call, but the answer isn't as clear for a short stop. There might well be some time during which the open-window car heats up faster as warm air from outside gets in, while the closed-window car holds in the air-conditioned goodness longer.

It occurred to me not long after posting that, while walking through the parking lot, that it was possible to test this with SCIENCE! A colleague in computer science has a car that is very nearly identical to mine, as you can see in this picture:


Chris's car is similar enough to mine that I have more than once stood next to it like an idiot pressing the "unlock" button on my car remote and wondering why the door wasn't opening. This makes for a nearly ideal test of the question of how to arrange your windows: just park both cars in the sun, one with the windows open and one with the windows closed, and monitor their temperatures over half an hour or so.

It took a while to get together, but yesterday, we did just that.

Here are the cars in the experimental configuration:


Chris's car is on the left, my car is on the right. As you can see, they're not quite identical-- the interior of his car is a light tan, while mine is a light grey. The extedror color is the same, though, and the two cars are the same model year.

i-65694e8eca18b91fbcddf31715b75e65-sm_thermometer.jpgThe high-tech temperature monitor can be seen at right. This is an indoor-outdoor thermometer, purchased at Lowe's. I bought two identical thermometers, and we set them up with the "outdoor" probe hanging halfway down the center back seat, so it was not in the direct sunlight, and the wire was run out through the door of the car, so we could read the temperature outside. These read to the nearest 0.1 degrees Celsius, and refresh the display every ten seconds.

The cars were parked facing north (more or less), and the test started at around 3:45 in the afternoon, so the sun was more or less due west. Both cars were in direct sunlight, and we took care not to cast shadows on the cars. We ran the air conditioners in both cars for almost half an hour, then I opened the windows of my car roughly 1cm each, and we left Chris's car closed up. We turned off both cars, exited quickly, then recorded the temperature once a minute for half an hour. Here are the results:


As you can see, there's an obvious flaw with this test, namely that the air conditioning in my car works better than Chris's. His car started about two degrees hotter than mine; the gap was closing a bit, but there's a limit to how long we were willing to stand around in a parking lot with our car engines running.

The trend is pretty clear, though: My car, with the windows open, started out two degrees colder than Chris's, but rapidly caught up and passed his car-- within two minutes, the temperature in my car was higher than his. It took another 10-15 minutes for his car to catch up and pass mine, and they do show the expected long-term trend, with the closed-window car eventually reaching a hotter final temperature.

If you're looking for the extra dorkiness you expect from a physics blog, here's the same data with a couple of curve fits added:


The solid blue line is a double-exponential fit to the data, which corresponds to two different heating rates-- a fast one (time constant of 2.7 min) due to direct exchange of cool car air with warm outside air, and a slow one (14 min) due to solar heating of the air inside the car. The solid red line is a single-exponential fit to the closed-car data, giving a single heating rate (about 8 min), presumably due to solar heating. The single exponential fit isn't great, but SigmaPlot chokes on a double exponential fit to the same data, so this is all I've got. I wouldn't put too much stock in this one.

We did some control testing by putting both cars in deep shade (that's the picture at the top), and confirmed that they heat at more or less the same rate, and both come to just about the same temperature as the outside air. An attempt in partial shade was abandoned, as it was too difficult to get identical amounts of sun on the two cars. The sensors themselves are pretty close to identical-- they take a surprisingly long time to come to the same value when put in the same car (in three-ish minutes, the gap between them had only cut in half), but given enough time, they record the same temperature to within a tenth of a degree or so.

What's the take-home message from this? Well, the clear implication of the data is that for short trips, you're better off leaving the windows of your car closed than open, to hold in the cool air. If you're just running into the store for a few minutes, closed windows will be cooler than open windows. Beyond 10-15 minutes, though, the closed-window catches back up, and the difference is never all that large.

Obviously, this would be greatly improved by starting the two cars at identical temperatures. If I could think of a way for this to end with one of the cars exploding, I'd send it to Mythbusters as a story suggestion (Chris suggested sending it to Car Talk, too). For the moment, though, this is the state of the SCIENCE! on what you should do when you park your car on a hot day.

More like this

Love it! I've been wondering this a lot as of late. Of course, I usually drive with the windows down and the air on, but that's just because I'm evil. Well that and a little bit of claustrophobia, I think.

This was a great idea for an experiment, but a few more repetitions would make this even more convincing, especially if you switched which car was the one with the windows open. Averaging over several such repetitions would help overcome the fact that the air conditioning in one was better, and that there are different interior colours, and perhaps even that the different people shutting the cars down may have had different body temperatures.

Also, is there any explanation for the dip in the blue curve on the graph? Did the wind pick up a little bit, and push a bit of air through? This variation could mean that it would be much be better to leave the windows open a little bit more if you don't know quite how long you'll be (especially if you also have a car alarm that could let you know if someone is trying to reach in the window and unlock it). Or, if it was just a fluke puff of air, it would be averaged out over several repetitions.

Future research possibility, if you're setting up to do this test again anyways: try out those windshield blinds, and see if they really do reflect enough sun that it's worth the hassle of putting them in place.

As a Brit I have to admit that the notion of air-conditioning in cars just makes me giggle uncontrollably. The last time I actually owned a car we kept it cool by opening the windows and deploying sunshades and blinds. Wherever possible we parked it in the shade. I think you need to define your assumptions about the prevailing outside climate and the nature of the car's cooling system. A "hot day" to me, living in a large estuary just off the northern European coast (London) may not match the expectations of a central European, an Australian or a North American from a southern state, and a drier climate would further complicate the applicability of your results. Coastal effects, altitude and the like might also factor in.

But those are minor criticisms. The basic approach seems sound. If I still had a car I think I would be able to have a decent go at replicating your results. Thanks for that.

By Tony Sidaway (not verified) on 28 Jul 2010 #permalink

I live in Louisiana where it has been 100+ almost every day for a month, and I have noticed that my Camry is very well insulated and takes quite a while to heat up inside. If I leave the windows up, the cool air from the A/C stays trapped inside and I actually feel the difference right after getting in the car.

Nice, but you should repeat the experiment with the open/closed windows on the cars swapped.

I think your long-time conclusions are sound, but I would bet that if you drive around a lot your engine will be significantly hotter than it would be if it just sits still in the parking lot, so that both cars would likely warm up faster than indicated here, and the difference might be washed out for short times...

Evan @9: Chad specifies that they ran the air conditioners for half an hour before beginning the data taking. That's long enough to get any engine as hot as it would get under most ordinary driving conditions. True, if you do a lot of mountain driving or you take it up to 100 mi/h on the freeway, you'll get it hotter, but neither of those scenarios applies to running ordinary errands, at least where Chad lives.

I agree with the other posters who advocate a control experiment where you switch which car had the windows open and which had them closed, as that will correct for the initial starting temperature as well as the different interior colors.

By Eric Lund (not verified) on 28 Jul 2010 #permalink

I think Car Talk would love this. You should repeat it with the cars' roles reversed, though.

I definitely agree that more repetitions, and swapping roles, would be good. The experiment takes at least half an hour per run, though, and there's a limit to how much time I'm willing to spend standing in a hot parking lot writing down temperature readings.

What you really want for this sort of thing is an automatic temperature logger, that would record temperatures electronically while I sit in the shade and sip lemonade. Nobody around here seems to have a portable one, though.

Would also be interesting to know about the effects of tinted vs. untinted windows. And those ugly side-window shades that have cartoon pictures of barney the dinosaur, or whatever. Oh, and those reflective sun shades that people put behind the windshield.

You do not mention wind speed and direction. It is likely this might be a big contributor to the differences in heating up in the interior

You do not mention wind speed and direction. It is likely this might be a big contributor to the differences in heating up in the interior

There was an intermittent light breeze from the west. The air wasn't completely still, but nearly so.

You can do automatic temperature logging with a Lego Mindstorms.

I've often done this, if I run into the store for a quick errand I'll leave the windows closed. Good to have numeric confirmation of this!

By Benton Jackson (not verified) on 28 Jul 2010 #permalink

I wonder if the cars were so close in temperature at the end because 1cm is relatively very little to leave the windows open? I'd be curious if you could try again, letting a bit more air circulate?

I usually leave mine open about an inch each and leave the sunroof vent open, too, and my car stays relatively cool in sunny Southern California.

By Lauren McAuliffe (not verified) on 28 Jul 2010 #permalink

What's the take-home message from this?

Bloggers are enormous dorks.

Isn't it an open invitation to thieves to a car with an open window?

Sili: That is a strong function of where your car is parked. Chad lives in a relatively low crime area, so he can get away with this. I can't speak for Chad, but I wouldn't do this in Newark or Los Angeles, or for that matter Manchester, which passes for a big city here in northern New England. I also wouldn't do this if I were parking the car at the airport or bus depot for a week, but for a different reason: too much risk that it will rain while I'm gone.

There is also a reason why the windows are only open 1 cm or so. A really determined thief could probably fish with a wire hanger and open the lock, but it would be easier to simply break the window.

By Eric Lund (not verified) on 29 Jul 2010 #permalink

I imagine others made the same comments, but were the thermometers swapped on a second test, or their calibration confirmed by some other means? This immediately occurred to me relative to the comment about AC effectiveness.

Also, the convective heat loss/gain is enormously influenced by wind speed. So actually, you could develop a family of curves showing heat loss as a function of windspeed.

Still, nice to see what seems intuitively obvious proved out. Convective heat gain through open window trumps conductive heat gain through vehicle envelope when the interior temperature is below ambient. For a while, before radiant heat transfer becomes significant.

As the interior temperature exceeds ambient (thanks to the radiant heat gains), convective heat loss exceeds conductive heat loss.

And so as we all know, if your house is cold, seal up leaks before adding insulation. If I may extend your findings.

I think you are getting only a very small effect from a window open only 1 cm. Based on nonscientific experience, I think they should be open at least an inch. Obviously the effect would be greatest if they were completely open, but you have to balance air flow and unauthorized access. Basically, though, if the window is open far enough to do much good, a determined person can get in the car. So the decision becomes whether you feel safe enough to leave the windows open at all. If you do, then open them more. If you don't, then put up a sunshade and leave the windows closed.

If you loaded the cars with explosives and threw in a heat-triggered fuse, you could probably do the mythbusters thing.

There must be a story out there about someone who brought home propane for the BBQ, left it in the car, and then it exploded....perfect for a debunking (or verifying, as the case may be)

By quietmarc (not verified) on 29 Jul 2010 #permalink

Y'know, this problem is screaming for an Arduino. Maybe someone (maybe even I) could wire up an Arduino to log temperature.

If we really want to get fancy, we can have an Arduino that logs temperature (interior and exterior), along with window states (somehow)...

What was the approximate outdoor temperature when you did this test?

The outside temperature (on the ground, in the shade of my car) was 36.8 C, and is indicated by the horizontal dashed line on the graph.

Two weeks ago we enjoyed temperatures of 40C and I was obsessed with keeping the house cool, so I enjoyed reading about you tackling a similar problem.

It's interesting to see the difference between guessing what might work, doing a rough experiment, and all the suggestions for improving the rigor of the experiment. That said, my guess is to agree with Tony Sidaway @3. If the idea is to keep the car cool, do as you suggest but with the blinds and sunscreens. Actually, I would love to know whether slinging a blanket over the air conditioned car is better or worse than a bit of airflow from a window. Maybe you could add that to the long list of possibilities for further experiment?!

A friend once (accidentally) left a bottle of olive oil on the seat of her car at the airport during a hot week, which exploded before she got back.
Now that's something which Mythbusters could have fun with...

By Craig Heinke (not verified) on 01 Aug 2010 #permalink

I don't think that cracking the windows would have prevented that unopened can of Coca-Cola from exploding in my Subaru. I bought it with me for lunch and left it in the car by mistake. :(

I live in the tropics and we here know enough to crack a couple of windows if parked in the sun to avoid cracking dashboards and windshields and split seams in seats.

Don't ever remember having to conduct scientific experiments to know this. It is instinctual, like knowing if you have to pee, where it is going to come out from.