When it comes up in casual conversation that I do not generally heat or cool my house, people either move to another seat or look at me with some mixture of admiration and disbelief. When non-Californians then find out that I live in San Diego, they might huff or spew, which often involves some embarrassing projectile escaping their mouth. But the locals are more consistently impressed—more so by my forsaking heat than AC (San Diego has very mild summers by U.S. standards). This summer, I turned on the AC for the first time since we bought the house three years ago. All in the name of science! I was blown away. Here is what I learned.

A Happy Convergence

I occasionally embark on a campaign to track temperatures throughout my house in order to understand how it behaves thermally in response to fluctuations in sunlight and external temperature. I keep telling myself that I’m establishing a baseline against which to compare the results of a hypothetical insulation upgrade/replacement. So far, I keep getting baseline data.

Most telling are campaigns when the external temperature is extreme in some way. The forecast showed a hot week was coming up, so I deployed the Thermochrons for an 8.5 day mission, set up to sample temperatures to 0.06°C precision every three minutes (and roughly synchronized with each other).

As it happened, the first day into the campaign was unusually hot and muggy for San Diego (high 30 °C, or 86 °F, 62% humidity). And we had invited 2500 W over for a surprise party (non-physicists might instead say 25 people). Okay, so Thermochron campaign, hot and muggy, loads of people coming over…if there was ever a right time to fire up our AC, this was it!

We Have Liftoff!

I set the thermostat to 75°F (24°C) flipped the switch at 4PM, and immediately rushed to TED to see what he thought of our unfamiliar visitor. I had never seen anything like it. TED was screaming that we had over five kilowatts running! Previously, the highest rate of electricity consumption I had ever seen in our household was in the neighborhood of 3 kW when both the microwave and toaster oven were in simultaneous operation.

I rushed outside to watch the synthetic dial “spin” on my electricity meter. Usually, I have to wait a full minute to see one block change, but now blocks were coming and going faster than once per second. Night and day difference!

100 W at a time, our guests arrived and were appreciative of the cool—though those who knew me well admitted to being surprised. The air conditioner ran constantly for 4.5 hours, and I began to wonder if the system had enough oomph to overcome the party load. But it finally started cycling, spending about 45% of the time in the on-state over the next five hours.

I clocked 32.6 kWh that day. This may not sound like a lot. In fact, it’s pretty close to the U.S. average daily household electricity consumption. But for me, it was out of this world. For perspective, over the previous 600 days, our typical daily expenditure (median) was 1.64 kWh (average 1.82, with 0.79 standard deviation). The minimum was 0.85, and the max soared to 6.3. 32.6 kWh beat our previous maximum by a factor of five, and our typical usage by a factor of 20! Welcome to America, my friend.

Daily energy use surrounding the air conditioning test. The test stands out like a huge, sore thumb. The vertical scale is kilowatt-hours.

Are We There Yet?

As much as I learned from this experience, the day of the “test” could hardly be called typical. Lots of people, lots of in-and-out traffic, a late start turning on the AC, etc. So I decided to dedicate the next day to a more stable test. Setting the thermostat at 76°F (24.4°C), the compressor kicked into action just before 10AM, executing 24 cycles over the next 11 hours.

The funny thing is, it felt unnaturally cold in the house, even set at 76 °F. My bare feet, dangling over the couch arm as I prepared a blog post, were on the verge of feeling cold.

Okay, a few anecdotes from “normal” America. I grew up in the South. It’s hot and humid in the summer, the humidity keeping it relatively warm even overnight. AC is ubiquitous, and I get this. But the thermostats are usually turned down so far that if I wear sandals indoors for any length of time, my feet get downright cold. No one in the entire state should have cold feet in the summer! I’d call it overcompensation. It really strikes me as absurd—especially in light of the stiff energy cost, and complaints about the swollen electricity bills.

In a second instance, my wife and I were visiting relatives in eastern Washington where it gets pretty hot in the summer and pretty cold in winter. We noticed the winter-time thermostat turned to 76°. Pretty toasty! When queried, we were told that so-and-so’s joints have difficulty if it’s any cooler than that in the house. Okay, bummer. A return visit in the summer found the thermostat at 72°. Yet so-and-so was moving around just fine, thank you very much. Maybe there’s some complicated medical thing going on here, but it sure seems to me that they likes their comfort, and that it’s largely psychological: they want to feel that home is a luxurious retreat from the nasty outdoors. Wearing shorts inside in winter and a sweater inside in summer somehow feels right as rain.

So here we are during our day-long “test,” suffering the unnatural cold of 76° in the summer. My wife asked me several times if we could just turn it off. “Aren’t you done yet?” I was a bit surprised by this. I’m usually hardier in terms of adapting to too-warm or too-cold conditions. What might have been considered a special treat had turned into a grueling march.

Adding to the trial, on the same day the New York Times Sunday Review had a fascinating article on air conditioning and its energy/environmental cost—to which I was now contributing. It also felt very strange to read about reluctance among many to bump up their thermostats to 75° or higher, while I was itching to ditch 76° as unpleasantly, unnaturally cold. Noting the outsized energy demand imposed by air conditioning, the article shared the scary statistic that if Mumbai, India adopted universal air conditioning for its inhabitants at American standards, the associated energy demand from this city alone would come to a quarter of the cooling demand for the entire U.S. That’s scary, folks.

Evaluating the Damage

Okay. Experiment Complete. What’s the damage? The air conditioning ran for about 400 minutes that day (nearly seven hours out of eleven). In this time, the air conditioning racked up 31.5 kWh of energy use. Over a 24-hour period, this averages to about 1300 W of continuous power. Imagine my fright: two days in a row over 30 kWh: more electrical energy in two days than we normally expend in a month!

So what do the Thermochrons have to say? Below is a plot of the entire period for the bedroom, which was not as disturbed by the party (not that kind of party).

Temperature record for the bedroom (blue) and outside (black). A cooling trend is evident, as is the impact of the air conditioner abruptly pulling down the internal temperature starting late in the first day. The fourth day (234) is unusual, in that I disabled the attic fan for the day.

If we plot the second and third days on top of each other (roughly similar outside), we can examine the impact of the air conditioner.

24 hours of AC experiment, compared to the following, normal day.

Now we can see that the day on which the air conditioner was run (the “AC day”) was on the whole a bit warmer outside (black line) than the following day (gray line)—especially in the evening and night-time. Meanwhile, the cyan curve indicates what happened inside on the normal (no AC) day, which turns into something like the dark blue curve when the AC is turned on.

One possible clue as to why my wife and I were unsettled by the AC—feeling that we had created an unnatural environment for ourselves—may be seen in the jagged temperature profile resulting from the cooling cycles. We were being whip-sawed around all day. I think this unfamiliar jerking behavior grated on us, getting progressively more annoying.

Quantitatively, I can say that the AC day averaged 1.07 °C hotter than the following day, over a 24 hour period (or 1.3 °C over 12 hours when the AC was active). Meanwhile, the air inside averaged 1.88 °C cooler inside on the AC day than the following day (3.5 °C cooler during active AC time).

Assuming that a warmer day outside translates to a comparably warmer day inside, we can conclude that the air conditioner bought us an average of 3.0 °C of cooling (1.07 plus 1.88) compared to what would have happened without AC, when averaged over the day (4.8 °C for AC-on time).

We can get an energy cost per degree by taking the power expended and dividing by “comfort” delivered. For the 24-hour period, this is 31.5 kWh divided by 24 hours, divided by 3 °C, yielding about 440 W/°C. For the shorter 12-hour period, we get about 550 W/°C.

What does this mean? Is it good? Is it bad?

Comparison to Fans

A box fan might typically run at about 90 W and move something like 2000 cubic feet of air per minute (cfm), or about a cubic meter per second. Imagine enveloping the fan within a window to efficiently move air between inside and outside (pushing either direction). Roughly speaking, if the replacement air is 1 °C cooler than the interior air, then each second, the kilogram (cubic meter) of cooler air removes about 1000 J of thermal energy from the house (specific heat capacity of air is about 1000 J/kg/°C). All that for 90 J of input, each second: a leverage of 11×. Good bargain. Better than a heat pump.

Looking at the graph above, the overnight house temperature stayed about 4 °C warmer than outside through the night on the normal day (cyan vs. gray). If a fan operating overnight (10 hours at 90 W) could even take that down to a 3 ºC difference, then this 1 °C bonus may average to something like 0.5 °C over the full 24-hour day. At a 24-hour average fan power of 37 W, this comes to 75 W/°C. Again, this would be a substantial bargain, compared to our 440 W/°C calculated for the AC system.

How much ΔT might we expect a fan to deliver? In terms of volume flow, it looks easy. At 2000 cfm (~1 m³/s), a 2000 ft² house (185 m²) with eight foot ceilings could have a complete air exchange in 8 minutes—if air could be coaxed into such an orderly flow. Of course it can’t, and some cool air will escape out of the other open windows. But in a matter of an hour or so, it should be possible to manage complete air exchange.

Yet experience says it’s not that easy. That’s because all the items and walls in your house are storing thermal energy, and donate this heat to any cool air that happens by, warming it back up. Every kilogram of material in your house (including interior walls and some fraction of the exterior walls) takes another roughly 1000 J to cool down by 1 °C, gobbling up another second’s worth of air using the foregoing numbers. So in practice, a single box fan will have difficulty moving enough air through a large house to substantially take down its average temperature. But what effect it does have will be a total bargain. Especially, blowing cool air into an occupied room is a good idea. And a beefier air exchange system can get some serious cooling done at a much lower energy cost per delivered degree than AC. But this mode of cooling is not available until the night cools off, so a well-insulated house is key to help preserve the night-time coolness for the rest of the coming day.

Comparison to Heating

Some time ago, I evaluated how much energy it took to heat my home in a similar “special day” experiment—this time as a Christmas treat. Using the construct of degree-days, I computed a cost of 610 W/°C for heating. Compared to this number, 440 W/°C for cooling is only marginally better, while I expected the energy efficiency ratio (EER) of the heat pump to exceed the direct-energy heating value by a factor of three or so.

But I’m not comparing apples to apples. Revisiting the heating experiment data, we generate the following plot for the heating day and the (similar) day after.

Temperature record for heating experiment. Inside temperature is recorded in our hallway near the ceiling.

In this case, the day on which heating was applied (black curve) was warmer than the comparison day (gray curve) by 1.6 °C. As a result of heating, the interior temperature averaged 4.5 °C warmer than it was on the comparison day (and almost certainly an overestimate based on the sensor’s location near the ceiling). If we apply our “warmer out = warmer in” rule, we only really got 2.9 °C of warming impact over the expected passive behavior (4.5 minus 1.6).

Doing so cost 3.5 Therms from natural gas, or an astounding 102 kWh of energy. I compute 1465 W/°C, blowing the AC cost out of the water by a factor of 3.3. Hey! There’s the EER I was looking for!

So the same house is less energy-intensive to cool via heat pump technology than to heat by direct flame. But let’s not forget that my utility electricity is largely derived from fossil-fuels, driving heat engines at maybe 35% efficiency. So in terms of primary energy use, every degree of environmental conditioning costs about the same for me.

Winter/Summer Aside

An interesting feature emerges when comparing temperatures in our house in the summer vs. winter. In winter, the interior temperature tends to be about 5 °C warmer than the outside temperature, on average. In summer, this is more like 2–3 °C warmer inside. I attribute the difference to our use of an attic fan during the summer (usually PV-powered), opening the windows at night, closing curtains during the day, and a different solar loading profile (eaves keep light out of windows during mid-day in summer; more summer loading onto roof and less into windows and walls).

Back to Normal

Our experiment is over. On my way home from the bus stop in the evenings, I pass houses with their AC on. But around that time, the outside temperature is dropping, so that opening up the house makes for a pleasant temperature. Granted, I’m in San Diego, where things aren’t so tough. Yet people did used to live without AC (and still do). Houses were built for it, with many windows allowing cross-ventilation. Those practices have faded, as AC is a feature taken for granted in the design of new houses.

I was walking through Washington D.C. one afternoon in late summer, and saw some old government buildings whose windows were propped open. “Must be a little stuffy and warm inside,” I thought, compared to modern buildings. The people working within may abhor their old building with its inadequate cooling capability. But then I considered a future scenario in which energy scarcity may impose restrictions on how much is available for air conditioning. In such a case, having a building whose windows can open will suddenly seem like a huge bonus. A building built to be tolerable in the days before AC will fare better in a future deprived of that luxury, should such a time come. The directional flow of envy between modern and older buildings may reverse.

For my wife and myself, seeing our daily utility electricity consumption shoot up by a factor of 20 to keep our interior cool (though still warm by American standards) was completely off-putting. It’s not the money spent so much as it is the principle of the matter. It feels like an unnecessary luxury—like having a servant peeling grapes for us. Humans evolved to withstand hot summers and cold winters. We’re tough stock.

Now that I have personal experience with the amount of energy involved in running the AC, I’m both more frightened and more optimistic. The fright comes from the fact that more people in the world strive to live in air-conditioned comfort—and I don’t blame them. But it’s hard enough as it is to meet our energy needs. Add more demand, and the strain is greater. To the extent that air conditioning is a necessary component of advanced society, we may have a hard time paving our way to that future, globally.

The optimism comes from seeing how much low-hanging fruit there is to lop off, if people are willing to change their thermostats to moderate settings (like 25 °C, or 77 °F, or even higher). Then another part of me says I have too much faith in our fellow citizens to adopt austerity for the common good. Not sure when I’ve seen that happen, en masse. I can’t do this alone, after all.