Monday, September 30, 2019

Sense Home Energy Monitor



A couple years ago I watched the below "This Old House" episode about a local company that developed a whole home energy monitor that uses machine learning to figure out how much electricity your home is using.  Basically every device in your home has a unique electric signature that sense can pull out and identify.  It is pretty cool technology that is still fairly new and while not perfect should get better as more and more people start using the device.  As a owner of solar panels I have a device that tells me how much solar I am generating but I have no idea how much power I am using at a given time.  To be able to drill into individual appliances may even allow me to change some habits and save some money.


The monitor is a pretty expensive piece of technology and add another $50 if you want to track solar output as well.   Not to mention it installs inside your electrical panel so you will probably need to hire an electrician to get it installed.

I had the monitor installed in mid-July.  It takes a couple days to start tracking your energy and then a another week or so before it starts finding devices.  My home is a bit unique in that my mother lives with me in an in-law apartment which means I have two stoves, refrigerators. ovens, toasters, microwaves and washers and dryers.  Not to mention two AC compressors which are my major energy users.  My main goal in purchasing the device was to get a handle on where I am using electricity but also as I tech geek I just think the technology is cool and fun to plat with.

So after about a week of having the device installed it started finding devices.    The "device discovery" process is not an exact science.  Usually when it finds a device it will tell you it found a motor or a heating device and a confidence level.  You need to go around your house and turn those devices on and off and confirm what it found then you can rename the device and provide more information to help sense.  Sense did a pretty good job identifying my devices properly.  It did find what it thought was three different vacuums none of which were vacuums that I own.   I think some of the oddities that I saw were due to a remodeling project I had going on in my house, so Sense found a bunch of power tools that aren't really part of my home.



So after about months of discovery Sense has found about 26 devices.  It has found most of my major appliances.  It will not yet find any LED light bulbs that you have but it did find the incandescent bulbs in my refrigerator.  It found my gas dryer, but not my washer yet.  It also has not found any of the TVs in my house all of which are LED.  Device discovery has slowed down over the past few weeks but I assume it will find a few more devices once the weather get cooler and I start needing to heat my home.



One interesting data point that sense provides is your "always on" or phantom power.  This is the electricity used by devices like your tv, cable box, routers, phone chargers or any other devices that use a small amount of power but are always on.  My always on number is under 250 watts which I think is pretty good considering the number of smart home products I have.   This is probably and area that I can take a closer look at and save a little money by unplugging some of those always on devices while I am not using them.


The device does integrate with some smart televisions but not any of the ones I own.  The also did recently announce integration with TP-Link smart switches and power strips.  So if there low power  devices in your home that Sense cannot detect you can plug them into the TP-Link device and name the device and it will show up in sense with its power usage.   I theory this should help sense detect some of these devices in the future.  I really wish Sense would provide a discount on these TP-Link smart switches as I would love to plug my entertainment center devices into smart power switch but for $70 I may wait for a sale.

The device is definitly a work in progess.  I have had much better luck with device discovery than most.  A friend installed the device a month or two before I did has less than 10 devices detected.  In theory though as more people use the device the machine learning models should get better.  I do have a concern about what happens if Sense goes under.  Since all of the processing is done on teh cloud, if the company goes away so does all of my data and monitoring.  But I guess that is a risk we take as early adopters of technology