The itemized electric bill

W. David Stephenson, a homeland security and disaster management strategist who shares my interest in citizen use of government data, pointed me to this item which suggests that if data produced by smart electric meters were shared in social networks, we could work together to optimize our energy use.

Sounds great. But as I understand it, although these meters report usage in hourly or even 15-minute intervals — thereby enabling utilities to fine-tune pricing and customers to fine-tune usage — they still can’t itemize your bill on a per-appliance basis. That would be a killer application. Imagine a little device that sits between the appliance’s plug and the wall socket, measures the power use, and reports that data over the AC network to a collector. Each device would be coded, you’d map the codes to appliances (TV, refrigerator, toaster, computer), and you’d wind up with a fully itemized accounting of where all the power goes. No guessing about the payback period for a new and more efficient refrigerator, you’d just know. A few years down the road, if your new Energy Star fridge starts to leak, you’ll be alerted to the fact and know to check the seals.

In this scenario the network effects would get really interesting. When contemplating the purchase of that new fridge, for example, you could go beyond the rated performance to the actual performance as measured by other users of that model. And maybe even adjust for factors like the number of kids in the house who are likely to stand in front of the open fridge door pondering their options.

As Amory Lovins points out, you can’t manage what you can’t measure. His MAP/Ming lectures on power use in industry are full of stories about unmeasured energy flows that, when instrumented, yield easy yet dramatic optimizations.

In software, of course, we know this instinctively. To optimize code, we inject instrumentation that shows us the hot spots where programs spend inordinate amounts of time. We need to inject the same kind of instrumentation into the electrical devices in our homes. I’m no engineer so I’m just asking: Is there conceivably a cheap, low-tech, easily-installed way to do it?


  1. Whoa, funny — I just ran into the inverse scenario yeserday. A customer was asking how to profile something (render times of specific parts of a web page in various browsers) for which there was no possibility of accurately instrumenting or profiling. I suggested doing a variation of A/B testing — take out pieces and measure; put pieces back in and measure; remove other pieces and measure. This is basically how we meaure success of UX enhancements to ad campaigns, and is the traditional way of measuring electrical consumption per appliance.

    Of course, the traditional method of taking a baseline per large appliance leaves some room for error and is not as precise as full instrumentation. But it’s not that bad, and you can easily tell whether you are in the right ballpark by comparing sums at the end of a period.

    I notice the same about tracking financial spend. It’s easy to get very detailed records of what you spent all your money on, but most people don’t know in broad buckets how much they spend on each major category. What % of net income goes to food vs. travel vs. housing? And so on. To have an accurate sense of these categories does not require precise instrumentation — you can instead do some A/B testing and focus on the parts that matter most, rather than tracking the details. For example, “I’ll increase my savings by x% for 6 months and see what happens to the rest”. Same as saying, “if I double the size of this table, what happens to page render time”? Or, “if I shut off the regrigerator for 1 week, what happens to the electrical bill”?

  2. Jon,

    You can use a to measure how much an individual appliance (or powerstrip) is pulling. You asked for cheap: they run about $25.

    “Connect your appliances into the Kill A Watt, and assess how efficient they are. A large LCD display counts consumption by the Kilowatt-hour just like utility companies. You can figure out your electrical expenses by the hour, day, week, month, even an entire year. Monitor the quality of your power by displaying Voltage, Line Frequency, and Power Factor”

    Of course, I’d love to see a wifi one invented that uploaded stats to your computer and had an embedded web server but I think the pricepoint would be rough :)

    – Mark Woodman,

  3. “if I shut off the refrigerator for 1 week, what happens to the electrical bill”

    And to the perishables :-)

    But, point well taken. In a smart meter scenario you wouldn’t need to wait a week, maybe just an hour.

  4. Of course, in the instance of the refrigerator* its power consumption depends on the surrounding climate and usage which would make it s consumption fluctuate hourly, daily, monthly and possibly even yearly. So a reasonable average would need to be taken over time to really see its usage. :)

    I do agree that a lot of energy usage is an enigmatic black box, but I wonder if anyone really would do anything about it even if they understood. It seems to me that a lot of people right now are interested in saving a questionable few watts per month by unplugging AC->DC convertors for cell phones and such all while gleefully staring at their 200Watt*12hr*24days Christmas Tree light display. They know where their big usage is, but they just don’t want to try to change it. I dunno… I love data, so I would be thrilled to have that sort of itemized usage but damned if I don’t see any but us geeky few really paying attention to it.

    To whit with Josh, back before I owned a Kill-A-Watt I used to do exactly what he proposes. I’d watch my meter and record average usage over a few minutes, run back inside and turn the item in question off (TV and computer were my major sources of curiosity) and then run back out to check the changes. It was a pain in the ass and had questionable results, since measuring the rate of a spinning analog wheel in the freezing Minnesota cold isn’t exactly what I’d call “precision”.

    So, I think that few people would really be concerned by the results, but damn it would make some interesting data for the rest of us to read.

    * Why is it called a REfrigerator anyway? Once the perishable item has lost its frigeration we tend to put it into the garbage or immediately into our stomachs… not back into the refrigerator…

  5. You could probably do this sort of data gathering right now by taking a kill-a-watt or similiar power meter and adding an interface to send data over the AC network, or RS-232, or ZigBee, or whatever other circuit you wanted to hack into the meter. How high is your “cheap” threshold? Assuming you can buy the meters for ~$30, and modify them for data gathering for another ~$30 – would it be worth ~$60/outlet to track your power consumption? I think you’d have to wait a long time before they paid for themselves…. You would be better off to have a few meters and use ’em to get a baseline for your devices, rechecking every so often. (Or just use an unmodified meter and manually enter the data for one device at a time.)

    Hmm… Google turns up which details metering of individual circuits in a breaker panel. You could extend the method to individual outlets, but it wouldn’t be cheap! However, I could see this sort of thing being done in new construction without breaking the budget – look how much money goes into buildings with smart lighting and HVAC.

  6. Whoa, kill-a-watt is cool! Speaking of using social networks to monitor things; Jeremy Cath has started doing this thing where he twitters his exercise log to a special account, and a web service snarfs the results for plotting. I think he mentioned doing this for gas fill-ups, too.

    It’s actually a pretty relevant topic to our industry; electricity is the primary expense of big data centers now.

  7. ” I love data, so I would be thrilled to have that sort of itemized usage but damned if I don’t see any but us geeky few really paying attention to it.”

    Maybe so. What would get people’s attention, here in New England where heating oil has quadrupled over the past five years, would be an easy/reliable way to quantify heat leaks. Compared to that, it’s true, the 17W draw of a digital radio that doesn’t turn off when you think it does is only spare change.

    “Whoa, kill-a-watt is cool!”

    Indeed. Thanks to everyone who pointed it out.

    “I could see this sort of thing being done in new construction without breaking the budget.”

    Yes, especially as economies of scale kick in to make the devices really cheap. But that’s no consolation to most of us. Easy/cheap retrofits or new solutions are what we crave. Like WiFi in old houses that would otherwise be a nightmare to run cables through.

    On the insulation front, I’m dreaming about a genetically engineered cellulose that you inject into your under- or uninsulated walls at just a few points, that grows to fill them, and then goes inert.

    Sheesh. Can you tell we’re having a hard winter? And this is only day 1…

  8. It sounds like a great idea, especially if you can roll some control logic into the network too so you can remotely turn things on and off.

    Pondering the possibilities has ended up with me buying one of the Kill-a-watt units, and your post led me to look when they were on special offer – reduced from 27 GBP to 10 GBP – so thanks :-)

    And today, I came across via Dan Hill’s delicious stream. It looks like it would do what you’re after (if you overlook the fact that they don’t do a US version just yet…) – monitoring individual power outlets and communicating via Zigbee or Bluetooth.

  9. Jon, while there are many types of circuit-level monitoring devices (wire-in, plug-in, and so on), few of them are “connected”, and those that are seem cost prohibitive at present. A relatively simple solution (for both new construction and retrofit) would be to place the sensing and metering (and, at some point, control and regulation) directly in the circuit breaker/distribution box. For extremely granular tracaking, it might require some re-thinking as to how circuits are wired (e.g. don’t branch in the room, branch from the box), but even room level aggregation would be an improvement. Alternatively, if someone can create a simple receptacle replacement with sensing/metering/communication all baked into onboard silicon, the cost could be easily driven down to a few dollars per sensing location. The home wiring could be used as the data communication backbone (that’s old news, been there done that) or, if it can be done cheaply, something like Zigbee or even extended Bluetooth could be used to transmit the values. The data could then be provided via an internet-connected device, and the consumer could have sophisticated analytics provided via a hosted solution. The consumer could “tag” each subcircuit with a friendly name/device type, and this data could be used (with permission) for aggregate usage analysis. In fact, if the device type was fairly detailed, it would provide true energy usage analysis for different brands/models of devices.

    The ultimate end game should be that the devices themselves contain this silicon (along with communications capabilities, local intelligence for troubleshooting, optimization, and alerting, and other functions) in a “wired home”. Again, once the functionality is driven “close to the metal”, the costs go down exponentially.

  10. Pingback: Marcus Pontin

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s