A Wall Street Journal reporter recently did an analysis of their home power consumption, the resulting article making the rounds of the meme sites. It appeared just as I was eagerly awaiting my own P3 Kill A Watt's arrival in the mail (no free units, or Ferrari 5000 laptops, were received for that mention).
Household Power Usages
I ordered my Kill A Watt to answer a question I posed on here last year - "Will Power Consumption Burn Vista's Avalon Aero Graphics?" In that post I pondered whether the 3D acceleration of Vista would measurably (and perhaps significantly) increase the power consumption of the host PC: If millions of PCs started consuming an extra 50W to power 3D interfaces, you're into the territory of building nuclear reactors to support Windows Vista upgrades. I'll get to that later.
In his analysis, the author of the WSJ piece -- Jason Fry -- surveys the electricity consumption of various devices throughout his apartment, trying to determine if gadgets are really to blame for increasing power usage. He concludes by declaring that "gadgets don't deserve the real blame for our nation's mounting electric use".
Mr. Fry came to this conclusion after estimating that some household appliances (such as a clothes dryer), along with general lighting, use ungodly amounts of electricity compared to his observed gadget consumption measurements.
On that point he is unarguably correct: The US DOE has been publishing standard power usage statistics for decades, breaking down where power goes in an average American home (note: average meaning that technophile young households are balanced out by a large number of technology-free senior homes, though of course there are tech-loving seniors and tech-fearing young adults). In a recent outing the DOE found that, averaged out, refrigerators (13.7%), lighting (8%), clothes dryers (5.8%), and air conditioners (16%) were significant electricity culprits, with home electronics seemingly minor players.
What Mr. Fry fails to observe, however, is the general trends in household power consumption.
Refrigerators and air conditioners, for instance, have seen dramatic power usage improvements over the past couple of decades. Even a decade old refrigerator can be significantly less efficient than the latest model, gulping upwards of twice the electricity, three times as much if it's two decades old. Furnace blower motors have been moving to DC, significantly reducing their electricity consumption as well. Even clothes dryers have seen significant efficiency improvements, not to mention that many coffee makers are moving to a much more efficient thermal carafe design.
On the lighting front, compact fluorescents have stormed the market, and are now more common on retail shelves than their obsolete incandescent predecessors, delivering the same light with one quarter the power.
Home insulation standards have been improving by leaps and bounds, reducing heating and cooling demands considerably: Homes might be growing in size, but they can be heated and cooled with much less energy per square foot than a home of just a couple of decades ago, and of course lighting extra space is far cheaper.
Virtually every major appliance has seen moderate to dramatic power improvements over the past several decades. Of course on a average basis some of these gains have been wiped out (from a purely per capita average perspective) as these appliances have gotten more accessible, with more homes having central air, dishwashers, and so on, but this lifestyle improvement has thankfully come in concert with efficiency improvements.
Despite all of these improvements in the major electrical contributors, however, the average monthly household power consumption has risen, as Mr. Fry notes, from 793 kWh in 1990 to 938 kWh in 2006.
How can we explain this seeming disparity?
My wife and I have been slowly replacing our light bulbs with the new compact fluorescent (CF) variant ever since they first appeared as a novelty at the local Ikea. We've kept buying as they've gone from being bulbous, slow to turn on nuisances to being instant on, pleasing, low-heat, seemingly last-forever replacements.
We're now at the point where every single non-specialty bulb in the house is a CF.
Outside, the CF floodlights are controlled by timers and motion sensors, and stay on for minimal durations. Inside I try to keep unused lights off, an exercise that usually necessitates covertly following -- hiding in the overly-lit shadows-- behind my wife, turning off the lights she's left on (she's not quite as committed to the pursuit as I am).
Our refrigerator, clothes dryer (with moisture sensor ensuring it dries the minimal amount to achieve the job), and dishwasher (with cleanliness sensor, always run without the unnecessary dry cycle) are relative new models, consuming far less electricity than their predecessors.
We could probably save the Earth a little more if we didn't do one to two loads of dishes a day, and if we didn't wash a dozen loads of laundry a week, but hey, that's modern life with small children. These are luxuries of modern living that I'm going to clutch onto until the ocean is lapping at the door.
Our air conditioner is just about to be replaced with a high efficiency SEER 16+ unit, but even that won't make much of a dent given that we live in the Toronto area. While we have the occasional hot spell, during most of the summer you can be entirely comfortable just by intelligently opening and closing windows and shades at appropriate times (banking night time cold air can be very effective in this pursuit). This summer past I doubt we had the air conditioner on for more than a half dozen days. The summer before was far more taxing on the AC, but the extremely high heat was an exception...hopefully.
Our water heater is a natural gas unit, as is our furnace, the latter having a high-efficiency DC motor.
So where's the power going in our relatively small, suburban home?
Behold the power usage graph from a somewhat recent power bill.
The months (at the bottom of the graph) go from July 2004 to July 2006. The summer of 2005 was freakishly hot, and our son was a newborn at the time, so we used the A/C much more than normal.
On average I'd say we use about 22kWh per day during a non A/C day.
My main PC -- the one that I'd leave on pretty much around the clock -- is a rather dated affair (it's such a pain migrating to a new PC). Featuring an Athlon 3200+, 1GB RAM, a couple of relatively small hard drives, and an nvidia 6600GT video card, it's empirically relatively power efficient given that it survives quite happily on a rather anemic 300W power supply, in an era when many advice guides are pushing monster 450W+ units.
This PC -- the midtower alone -- uses 129W doing nothing of consequence, with only the idle thread counting cycles for fun, spiking to 140W when it's CPU saturated (such as running SETI@Home around the clock), jumping further to 165W when a demanding 3D application is running.
I often keep this PC on around the clock to allow me to access it remotely, and to make it immediately accessible when I have the need to jump on and do something. It is hardly uncommon in the technology industry to have an always-on household "server" of sorts.
Being kind (or unkind) and presuming it sits doing nothing, it's still eating through 3.096 kWh of power per day! Doing nothing. Here in Ontario that comes out to about $0.32 a day (given that the fully-priced incremental cost of electricity averages around $0.106 / kWh), but more important than the almost negligible run cost, it's a fair amount of resources being used perhaps unnecessarily.
To try to achieve a bit more conservation, I enabled Wake-On-LAN and started putting the PC in standby mode when I wasn't using it, sure that it was yielding great savings. Imagine my surprize when I discovered that the so-called standby mode still slurps 112W!
I thought I was saving the Earth, when actually I was just adding inconvenience for myself.
( I've since discovered that a quirk in XP had it not using the full ACPI features of my system, and have since started going to S3 standby, dropping the unused power consumption to a miserly 5W. This is the same power consumption the system uses when it's "off" )
Now add in the two 19" CRT monitors. I still use CRTs because they still work really well, and when purchased were high-end units, and still feature great display quality. I'm also realistic that the LCD power usage isn't nearly as good as often presumed, especially in the ultra-power backlight new units). Each of them uses 67W when displaying a black screen, peaking at 88W with a fully white screen (LCDs generally don't exhibit the same fluctuating power usage, as the backlight is generally on all the time at full brightness, though some new home theater LCD TVs can lower the brightness of the backlight to extend the dynamic light range for very dark scenes). A comparable LCD screen would use about 45-50W (yes, I've validated those numbers, and a 19" LCD really does use that much power. There is a going myth that LCDs are far more efficient than they really are).
Splitting the difference, that's 77W per monitor when not in a power save mode. 154W in displays, coupled with 129W+ of computer, coupled with a 5-10W speaker system. The printer/scanner uses marginal, infrequent power, so it isn't worth a mention.
Averaging out and very conservatively assuming that the computer gets 50 hours of active use per week, 50h*(135W+154W) = 14.45 kWh / week, otherwise acting as a listening server (118h * 129W) = 15.22 kWh / week, that accounts for a total of 29.67 kWh per week, or about 4.2 kWh per day.
19% of our power consumption going to one PC! This isn't even my primary PC, but instead is more of the star around which various other PCs and laptops revolve, a deployment model that is appearing in many tech-savvy households.
Add to that the 5W used by the speakers, the 7W used by the switch, the 10W used by the router, all on around the clock, and this stuff starts to really add up (another 0.5 kWh per day -- almost 3% of our total daily consumption -- going to a couple of unnoticed little devices).
20 years ago a home PC was a Commodore 64 or an Atari 800, or if you were wealthy an Atari ST or just released Amiga. Those personal computers of the day used around the same amount of power that my lowly router uses today (itself a little computer running Linux). Now PCs are power hungry maniacs, and we often have households with multiple PCs, several on around the clock.
Over the past decade the power consumption of home PCs has increased significantly, with brief retreats as processor generatures mature and see efficiency improvements. Game entertainment units have seen the same power-hungry march, with new units featuring massive external power bricks, often featuring frequent heat-related crashes.
How about watching television? Surely we've made great strides in television watching.
20 years ago you'd huddle around the 23" color television, and that was the entirety of your "home theater". Nowadays watching television often entails powering up the PVR, the receiver, and the big screen television. Our own television setup is extremely modest by regional standards, and is comprised of a standard 5.1 receiver, PVR, DVD player, and 27" LCD television. I'm going to simplify and ignore the media PC.
Powering up the units, we see that the receiver/audio amplifier takes 51W, regardless of output (if you really crank it you can get it to start to spike, however the difference between loud and really loud is only a W or two of power output, going to ear-piercing when you exceed that). Turning on the PVR we find that it takes 42W. Actually it takes 42W regardless of whether it's "on" or not, as it still sits there chuttering its hard drive and thinking about world domination when it's supposedly off. Powering on the TV, you can add another 110W, despite the fact that this is a "low-power" LCD unit, with a much smaller backlight than many of the larger, ultra-high-contrast units available.
203W just to watch television.
Say that it's on for 36 hours a week (like most families, it's often on even when unwatched, especially given some of the great digital music channels our cable provider streams out), and "off" for the other 132 hours, that means that it's consuming about 36h * (51W+42W+110W) + 132h * (42W...remember that always siphoning power PVR) = 7.3 kWh + 5.5 kWh = 12.8 kWh, or about 1.8 kWh per day. Another 9% of our power usage going to watching television, despite us theoretically having a lower power setup.
Our television power usage would be magnitudes worse if we had a large screen plasma.
Think ahead a bit, and contemplate what the endless decline in LCD prices will result in: it seems obvious that any power efficiency gains of LCDs over classic CRTs -- an efficiency improvement that's already been dwarfed by the appetite of accessory devices in a normal home theater -- will get wiped out as we move to larger and larger displays, with monster backlights.
It is entirely foreseeable that the near future will see a living room with multiple screens (similar to dual or tri-monitor computer setups that you see today).
29% of our entire household power usage going to an infrequently used PC and the pursuit of television.
It hardly seems like gadgets are quite as innocent as Mr. Fry made them out to be. I don't even need to add up the millions of tiny nibbles by battery chargers, accessory devices, and so on, to realize that they are a noticeable, discretionary contributor to the demand on the grid.
Fire up a PS3 on a big screen plasma and things would really get ugly.
Worse still, predications are that it's only going to get worse. The US DOE predicts that "Electricity consumption for home electronics, particularly for color TVs and computer equipment, is also forecast to grow significantly over the next two decades. EIA projects electricity consumption to grow 3.5 percent annually for color TVs and computer equipment through 2025, to more than double the level of consumption in 2003."
Ouch!
Unless you're shining lasers at alien planets, all of that wasted energy is realized as heat: The hotter the device, the more electricity it's consuming (and vice versa), regardless of whether the juice went directly and intentionally to heat -- as in a stove -- or as an unavoidable side effect of some other pursuit -- such as analyzing radio signals looking for alien chatter on your home PC (perhaps we should combine goals and the SETI@Home people can distribute new oven burners comprised of a grid of networked multicore processors. Win/Win!).
If you live in a hot climate, or in a moderate climate during heat spells, all of that heat is a bad thing: It means that it's doubly wasteful, as it was wasted energy in the first place, and then is additional load on the overtaxed A/C. We've all experienced walking into a room hosting a PC rig to find it unbearably hot, this computational space heater spilling out endless warmth while the rest of the house is cool.
If you live in a colder climate, however, that heat actually contributes to household heating, reducing the load on the main heating supply. If you use electric heat in your household, then all of the wasteful appliances really aren't "wasting" during months when you need to heat your home, and that PVR is just as efficient at turning 42W into heat as your baseboard would be with the same power. This presumes that the heat isn't excessive, and is where you want it.
The point is that it isn't all loss in all situations. I can say definitively that since we've put more effort into being energy efficient in hopes of saving the planet for future generations, our furnace comes on quite a bit more. Of course natural gas and a high efficiency furnace yield more BTUs per dollar than electricity, but it does undermine the savings just a bit, and it is a pragmatic consideration to keep in mind.
As mentioned at the outset, I originally ordered the Kill A Watt to get some solid numbers on whether Vista increases the power usage of a normal PC over its predecessor.
It doesn't.
The measured difference between doing general desktop tasks on Vista Ultimate running with Aero Glass, and Windows 2003 running on the same hardware, was negligible. I could cause power spikes if making unreasonable demands of some of the new Vista 3D features, but could only cause a measurable difference doing something ridiculous with the intention of causing a spike.
Having said that, unsurprizingly some of the 3D screensavers included with Vista cause a 20W power consumption increase as the graphics hardware is enlisted to help draw some spinning 3D text on the screen (years back I worked at a small engineering shop during a period when we proudly went from using someone's desktop as the "server", upgrading to a real, albeit low-end, Dell server. Frequently the server would seem to start choking, taking forever to perform simple tasks like transferring a file. The "sysadmin" would go up and take a look at the physical box -- Windows had no real remote administration tools at the time -- and declare that everything was fine. It did seem to be fixed, but then twenty minutes later the sluggish response returned yet again. The problem, as you probably guessed, was that the machine was configured to display a 3D text screensaver when not in use: given that the machine didn't even have hardware 3D acceleration, the CPU was saturating doing transform and lighting computations for a screen saver for a box sitting in a tiny closet unseen).