Comparing Home Power Analysis and Disaggregation Products

SenseLabs’s partner Schneider electric has also partnered with SustLabs to manufacture a strikingly similar device called OHM https://ohmassistant.com a home energy monitor that is compatible with India’s power grid. Which may be contributing to the confusion amoung Sense support employees.

That being said, OHM claims to have proprietary technology, but also uses AI (Machine Learning Bot) to natively detect devices, and has many similar features:
“OHM For Home Features
Real time Energy consumption in Watt and Units
Analyze the consumption trends
Appliance Level segregation - yesterday, weekly and monthly
Diagnose health of your appliances
Get Energy Statements
Electricity Bill forecast
Pay your electricity bill as well
Last 24hrs consumption pattern
Energy units graph weekly and monthly
Forecast electricity bills
Discover power leaks
Vyas - User asset management
Get notified sudden surge in consumption
Easy to use mobile app with an intuitive interface
Know which devices are functioning properly and which are not”

OHM has distinct advantages over Sense like 24/7 phone support. Oddly, many of the things OHM lacks are the same things most Sense users have been experiencing issues with such as Lab’s.
I cannot find any information regarding a direct partnership or licensing agreement between Sense and OHM, but that’s not to say one doesn’t exist or it wasn’t done through Schneider Electric acting as the middle man.

On a side note, OHM Connect (not to be confused with OHM) has it’s own partnership with Sense and Schneider and works primarily with utilities to directly control your home. Essentially you give them control of your IOT devices, allowing them to shut things off (like your A/C or Fridge) during peak usage times and they pay you.
OHM Connect is much more invasive than our utility’s program which offers $45 off and rebates on a Google Nest or Ecobee thermostat if you agree to allow them to control your AC in the summer.

2 Likes

Just took a quick look at the first Ohm you described. Interesting, but very different product.

  • Not a separate higher-resolution monitor - Uses the existing in-house meter if you read the fine print.
    • Subject to the limitations of the utility meter and possibly the utility’s reporting network.
  • Based on my experiences with similar devices, there are quite a few issues with that approach
    • Not enough resolution to do short term device power disaggregation/recognition (bubbles).
    • Might be able to do more general device recognition, over longer periods.
    • Reads only grid net usage for solar / battery installs
    • Can’t show Total Usage
  • India call number (+91) - probably not a great match for North American customers.

True, though in many ways it’s extremely similar.
The biggest difference is that OHM can natively detect devices similar to Sense, but also allows users to agree or disagree with detections. Agreement is confirmation, disagreement allows the users to input what they think a device is which causes the Ai (bot) to try harder with detection until it can be confirmed as accurate. Device detection data is more individualized to the user staying a bit separate from the AI until so many other confirm it’s accuracy. It then uses that to help train the AI’s ML process.
Sense on the other hand only allows for strict Ai detection or deletion (start all over) of a device with virtually no human input.
I can totally understand the reasoning why both methods were initially created.
The ‘can you train Sense’ question users have been asking about for years will be quite the quagmire for Sense if this ends up working (long term) for OHM.

  • Not a separate higher-resolution monitor - Uses the existing in-house meter if you read the fine print.
    It’s a separate device installed in a panel (din rails). I think it’s hardwired inline thus negating the need for CT’s.

  • Not enough resolution to do short term device power disaggregation/recognition (bubbles).
    We all do love our bubbles… But there is this…


    image

  • Might be able to do more general device recognition, over longer periods.
    About the same as Sense to start detecting.

  • Reads only grid net usage for solar / battery installs
    image

  • Can’t show Total Usage
    It does. It can even work with Home Assistant.
    image
    image

  • India call number (+91) - probably not a great match for North American customers.
    [/quote]

If OHM were to outsource support to the US where no one has ever used the product as it’s not compatible with the US grid, their customer base in India would be upset.
Sense on the other hand a product only compatible with the US type grids is outsourced to India where no one has ever used the product but no one at Sense thinks this is an issue.
With different types of energy grids and different technical terms than one is familiar with or understands it’s extremely easy to confuse and/or misinterpret lot of relevant and important details.

Note they use terms like “Sense-abled” and “Wiser-Home”

image

I’m thinking that based on Ohm’s description (“connects to your electricity meter”), the hardware is similar to this product, the Rainforest Automation 200, which I already own. There are two pieces to the Rainforest 200 product, the “box/hub/gateway” which essentially connects my PG&E meter to the internet, and the Web App / Mobile software which offers a very stripped down set of functionality compared to Sense.

Hardware
Unfortunately, as you can see on Amazon, they are no longer selling this product due to supply chain issues (they couldn’t get the the internal chips in small volumes, compared to autos, anymore).

Software App
Even though, this is what they show on their website:

This is what I actually see:

  • No devices (even though they claim to AI and disaggregation)
  • No solar - (importing solar via a file,as you say Ohm suggests, would be farcical endeavor)

Here’s their equivalent of the Power Meter - similar but:

  • No solar, hence the usage is Net Usage rather than Total Usage for the house
  • The update interval is 5 sec vs. 1/2 second with Sense. That’s because the power meter only updates energy data every 5 seconds.

1 Like

As for the “allows users to agree or disagree with detections” part. This sounds like a little bit of a repeat of the Neurio product, which seemed similar to Sense, except it focused on human guided training.

I deliberated about Neurio or Sense back when I first bought 4 years ago, but glad I bought Sense, because if you look for Neurio today you find it in three places:

  • Ancient YouTube videos of “training” and “installation”, just like I shared earlier.

  • Deep in the bowels of the Generac website - Generic bought them a couple ago, partially because their training-based detection really didn’t work. But try to find where it has gone to:

https://www.generac.com/for-homeowners

  • Amazon - you can still buy cheap Neurio kits seemingly forever, but who knows how long they will keep the back-end in place and the app working.

Well, I have a better one.

While you guys were going on about an Indian add on, Duke Energy just sent me an email with my 15 day (half month) usage on my electric bill and projected full month bill.

However, this month they broke it down by category/device.

My first thought was that Sense was selling my info to Duke Energy.

After I realized that’s not happening, I checked always on, cooling etc in Sense vs what Duke Energy stated in their email and they were within 10%.

While you say “they were still off by 10%”, my usage is always off by around 10% Sense vs Electric Company.

So here we have Sense with learning curves, around 30 KASA plugs, Hues bulb monitoring and me inputting what I have in my house vs Duke Energy with, well, nothing and getting it to within 10% of where Sense is reporting.

I’m now thinking, why bother :man_shrugging:t3:

1 Like

I have opted in to a similar service, Home Intel, from my utility, but it’s a far cry from Sense when you look under the hood (both good things and bad things). Here’s the web interface:

Positives

  • Straight from my electric meter - simple and aligned with my billing.
  • Combines electricity and natural gas usage - uses $$
  • Gives loads of (obvious) suggestions with estimated $$ impact

Negatives

  • Straight from my electric meter - has all the limits my meter imposes
    • No solar info - uses estimated solar for some calculations, but could be way off.
    • No leg by leg insights
  • Combines electricity and natural gas - means that all comparisons are done in $$
    • Hides actual energy usage and time of usage
    • Makes trade-offs harder due to poor visibility
  • Lowest resolution is Monthly / Billing Cycle
    • Can’t see the results of changes quickly
    • Can’t see problems without a long latency
    • Can’t do experiments like pulling the plug or flipping something off
    • No other resolutions
  • No integration with outside (Home Assistant) or data export
    • Would be pretty useless anyway, with 1 month updating
  • Seems even more prone to operation problems than Sense
    • I think they lumped my April into May, but I haven’t gotten an answer from them in 5 months !
  • Low Device resolution - lumped into big categories instead of specific devices (same as yours)
    • My heating and cooling estimates from Home Intel are about 25% off - hard to determine exactly because they don’t break out electric and and gas and is all in $$
  • Forces users to go through onerous questioning/survey about devices in their house
    • Drives users to answer multiple choice questions where none of the choices are applicable
    • Also invokes guilt - “That device isn’t a good choice !”
  • Just like Sense, it doesn’t use whether you drive 1, 2 or 3 EVs as part of its similar home comparison.

Pretty clear why I use Sense vs. this nascent technology. But some people would be happy with it.

ps: I do like their Energy Heatmap, but I wish that that they had more flexibility in time ranges and axes. Everything in their current heat map is fixed.

I haven’t answered any questions. ZERO about what was in my house. Just showed up as you see it.

And again it’s within 10% of what Sense is reporting.

In the last 12 hours since it came in I have now gone in answered a number of questions about my house. eg Heat Pump, Central Air or none. Also asks about additional window units. Number of Refridgerators and/or Freezer units in House, Dryer AC or gas, Oven AC or Gas, Stove AC or Gas, computer or not, lights % LED etc

I will be interested to see how close it hits to Sense numbers now with additional input.

And this showed up 1/2 way through bill cycle. I imagine that means it would produce at minimum 24 a year.

Sure, Sense has more granular data, but the real question is how many need that for $200+ dollars vs this for free.

Secondly, this has learned this much about my house with no additional devices or input from me, unlike Sense.

If Sense wants to be worth a Premium Price, then it should certainly be able to learn and detect with more accuracy and give me more info that the freebie does.

1 Like

The world keeps getting more and more interesting. This isn’t disaggregation, but a company that gives better access to power data. It builds a consistent API for all sorts of different utilities.

The developer API for energy data
Pelm’s API empowers developers to build applications that connect with utilities​ to harness energy data.

https://pelm.com/docs/reference/utilities#supported-functionality

But it also shows the challenge of systems that use the current utility infrastructure - all data arrives a day or 24 hours later and the best resolution is every 15 minutes.

https://pelm.com/docs/reference/utilities#supported-functionality

2 Likes