Using Sense to save energy

Thanks for your feedback. As I mentioned in my post, we are indeed continuing to focus on the basics – so mainly working to increase device detection and user feedback around this.

While the results on each of your houses may vary, we’ve almost doubled the average number of devices per house over the past 8 months. And, coverage is continuing to increase – we’ve been releasing model improvements on a continuous basis, and there are some bigger improvements coming soon.

I know a number of you have been asking about directly training Sense – I wrote about this in a blog post a while ago: device detection. As I mention in the post, we’ve tried direct user training in the past, but the main challenges for device detection are having the system learn whether it can correctly model something (being sure we are going to know when it turns off before showing you that it is on!) and whether there are competing device signals in your house (a pump in your dishwasher which happens to look like your sump pump). Getting labels from users doesn’t help with these two main challenges. The devices which show up (either named like “microwave” or unnamed like “motor 2”) are after the system has sorted out these two issues – that is when it is indeed helpful to get user input. So, it’s not that we don’t want your help training Sense, we just want to make sure we are doing it in ways which are actually helpful!

And, we do already rely on crowdsourcing – the labels you all give unnamed devices (so, remaining “motor 3” to “pool pump” for example) have been a major driver of our continued model improvement. In addition, we will be soon release more application features to allow you to do things like merge devices (if Sense finds multiple parts of your oven for example). This should not only allow you more control over how Sense is working in your house, but will continue to feed into better models across houses.

4 Likes

Just so the Sense team doesn’t think the forum is degenerating into a Facebook/Twitter level Lord of the Flies, I have confidence in the product. It’s the best product on the market now for my needs. I’d challenge those frustrated to find anything on the market which can do what Sense does at $300. If you’d have done your own research you’d have found that Sense allows for:

  • Accurately reporting real time and daily solar production and household usage ✓
  • The ability to identify appliances with just a pair of CTs ✓
  • A user friendly app ✓
  • Ongoing improvements and regular updates ✓

Is it perfect yet? No. Do they have a viable product that meets or exceeds features offered by the competition? Yes. Not happy? Return it and wait 5-10 years for a better home energy monitor to be developed. Sense’s failure is being a few years ahead of its time, on the cutting edge of a technology many of its purchasers do not understand.

13 Likes

Hi Mike,
We have had Sense installed for two months. It still has a long way to go but I’m interested in the technology and think it will succeed in the long run.
We had solar panels installed at roughly the same time as our Sense. We knew our array would be slightly oversized, but didn’t really know by how much. (Our monthly electricity bills were skewed for a variety of reasons). Because of Sense, we now have the data to justify taking some appliances from gas to electric in the future. (We were told this was a crazy idea).
We are also considering battery storage when net metering goes away in our state. I wouldn’t think of purchasing storage without this data. Sense shows how we can change our electricity use throughout the day to more closely follow what the array is producing and reduce storage requirements. We can instantly identify best to turn on the dryer or charge the electric car. I’m hoping that in the future, this will be automated…Sense will interact with smart devices/relays to direct excess solar power to the electric car or hot water heater, based on realtime production and appliance use. IFTTT integration is a great step - can we get more triggers please?
Yes, maybe these uses are edge cases, or maybe they’ll be mainstream in 5-10 years. But I can say that your great looking app is a nice talking point with other tech-minded people who are interested in reducing their own carbon footprint.

6 Likes

I’m satisfied with the purchase I’ve made thus far.
I’ve already saved money with the device. I haven’t shared my story here, but I will now…
I easily identified a clogged exhaust line for my dryer based on the very short heating cycles shown. After inspecting my dryer exhaust, I found a backup. So as far as I’m concerned, it saved me electricity usage and a service call to a dryer vent cleaning service.

I knew at the time of my purchase there would be bumps ahead, it’s something that occurs with every new product. That being said; I’ve had more problems with well established products far beyond what I’ve recently had with Sense.

There are many improvements that can be made, but that will come in time. Apple didn’t become successful in two years. I’m more excited for the fact that Sense uses our input and changes or adds things based on our opinions and reviews.

I can’t help but wonder if there’s a way to make older adopters capable of labeling? While I wouldn’t trust the majority of people labeling items, maybe there’s a way to trust a small percentage to speed up Detection? I read the blog and understand why the learning process is important. But I can’t help but think if something is labeled, shouldn’t it learn at a faster rate considering certain perimeters are absolute? Not to mention, labeled or not labeled, it’s still learning regardless.
I own and operate a Mechanical Company and I would love to sell these as a techy appliance, But I’m waiting for better learning & accuracy.

On a side note, I’m really impressed with the communication I’ve had from Ben & Customer Support. Obviously, I can’t speak for everyone, but I’m really satisfied with their reaction time.
Don’t get discouraged & Keep up the good work.:+1::grin:

5 Likes

I’m another person who is satisfied with the cost of Sense vs. what I have learned and continue to learn about electrical usage in my house. Admittedly I had fairly low expectations and was willing to gamble $300 on something that I wasn’t completely sure would work. OTOH I am in the tech business and understand how it’s possible for a product like Sense to improve over time and I have the patience to give it that time. Would I like Sense to have detected more devices in my house over the 5 months I’ve had it? Sure. Will it make me more appreciative when my next device gets detected? Absolutely, because I have a sense of how much work is behind the new detection.

I also have an idea of how much a company can (i.e. cannot) disclose without jeopardizing their IP and other proprietary processes. Even though from everything I’ve read Sense is the best device of its kind on the market, they do have competitors and we are kind of now in it with them if we want to protect our investment. So pushing to get information that the company can’t disclose is counterproductive. That said, maybe it’s not so bad for @BenAtSense to be authorized to say something can’t be disclosed, when appropriate, so we have a reason rather than leaving things vague.

Even before I ordered Sense I knew I was going to volunteer for the beta program and I did. Unfortunately, that seems to be limited to the web interface, which we know is on hold due to resource issues. I suggest that the beta program’s mission be expanded to disclose to a certain vetted group (and under formal NDA) a little more about future plans and where some of the bottlenecks are. I am really impressed with the knowledge that many of the contributing members of this forum have about electricity, principles of detection, how it all works, etc. We are a resource the company can tap in to.

8 Likes

I’m quite satisfied with Sense for three reasons:

  • Reasonable expectations - I’m an electrical engineer and I have a rational view of how hard the problems are:
    • Just accurately measuring AC split phase power with a couple of pairs of CT clamps. For those of you who say that this should be easy, you have never done a lab on measuring real and apparent power.
    • Making sense of household power waveforms, given the ranges of device operation, the cacophony of different devices, plus induced noise from power generation, harmonic interactions, etc.
    • Using machine learning and AI, especially with the limited CPU capabilities of the probe device.
  • Reasonable progress
    • I’m up to 28 active devices, most of them fairly accurate.
    • I’ve seen EV detection work fairly accurately though it seems to still come and go. I’ve had a couple weeks where over 2/3 of the power was traceable to detected devices (higher % if you include “Always On”)
  • Useful results, even without Sense detection
    • I have worked to understand and lower my “Always On” from 620W to 320W and know how to go lower if I need to. I also know when my son has left his gaming rig on overnight… Have also spotted and turned off some outside halogen lights that were left on overnight, in a location we can’t see directly (250 extra watts in Always On).
    • I have an accurate time view of actual usage vs. solar production, not just some net readings…
    • I can see what’s happening over time with the big energy hogs in my household, even without detection. I can identify car charging for all three cars (two EVs and one plugin hybrid), plus upstairs and downstairs AC, to adjust use behaviors if necessary.

I’m in it for the long haul but have already seen 300$ in value.

10 Likes

I agree. Sense is not perfect, but I’m willing to spend the $300 to see how it works. My frustration is that while it has found many obscure devices, in over 2 months it hasn’t figured out the vent fan in the bathroom or the bathroom heat lamps, both of which are used daily. These would also seem to be easy types of devices to identify. If it can’t get these in 2 months, when will it? Without a way to train devices, I can’t see how it would ever figure out the devices buried in unknown. I am bringing these up not to complain, but to point out that there are issues. If we don’t bring them up, the guys at Sense will never know there is a problem and will never fix them.

3 Likes

I’ll share 2 energy saving stories, both concerning my well pump.

After Sense detected the pump, I noticed that it was short cycling, turning on almost immediately after turning on
a faucet and then not running very long. That’s not good for the pump. Turns out that in the 10 years since I had replaced the pressure tank, all the air had leaked out of it. Pumping it back up to the proper air pressure solved the problem.

The second issue concerns water leaks. I noticed that the pump would turn on at random times, even when I wasnt home. I went around and fixed all the leaky shut offs such as the ancient valves in the toilets, hose shut offs, and drippy laundry tub faucets. Now the pump turns on only when demanded.

Granted, its not much of an electricity savings, it accounts for only 1% of my total electric bill (I didnt know this before I installed Sense), but I figure avoiding any unneeded wear and tear on the pump is saving me from a huge repair bill in the future.

5 Likes

Sense absolutely saved your well pump. Any motor that constantly short cycles will fail prematurely.
Bladder tanks are one of the most common causes for well pump failures.

Maybe there should be a section for how Sense saved some of us money or damage under the “share your stories” section.
I’m willing to bet there are quite a few stories out there.

4 Likes

Hi everyone,

Thanks for the input so far - these are very useful! And yes, please include things beyond just saving energy.

I do want to get someone working on fault detection at some point. We already see some sorts of faults in the signals (seized compressors, tripping breakers, etc.) but haven’t yet had a chance to look at these systematically. We are thinking that we will need to do specific modeling of particular things which can go wrong (like the clogged vent changing the cycle time of a clothes dryer) and will need to prioritize the list of things to go after.

So, letting us know what you are seeing is great and we will use to prioritize work once we have some cycles free to look at this.

Mike

4 Likes

I’ve been wondering if the failure of my water heater to be recognized repeatably (I deleted it a second time earlier this week for poor recognition) is because it may somehow be failing? It’s an old water heater, probably 10+ years - maybe as old as the house at close to 20yrs.

Anyway, the issue is that about half the time it would pick up both legs of runtime, the other half it would only pick up one leg (based on the power meter data, I can tell both are running). I bet if I went out and bought a new water heater it’d be detected better, but I’m not ready to consider that because it’s still functioning and not costing us anything because we’re full solar :slight_smile:

Hey @NJHaley, I assume you’re using a standard 240Vac electric water heater. Most likely a 40-60 gallon.
Most heating elements are approximately 4500 watts, so it should be easy to spot it on the graph assuming it’s working properly.
Contrary to popular belief, Only one element runs at a time. They switch based on thermostatic temperature. The upper element controls both and sometimes the bottom element is a smaller wattage. That being said, depending on the age of the water heater, wattages could definitely be affected if an element is on its way out. It may be beneficial to get a water heater tune up kit. It comes with the elements and thermostats needed and it’s fairly inexpensive. It can be a diy project but it’s usually best to have a professional do it.
If you do decide it’s something you want to try, feel free to contact me for any advice in doing so.

1 Like

Correct on all counts, @vrai.kenny. The curves I see are about 5kw. Sometimes Sense would see the whole 5kw, sometimes it would see just 2.5kw-ish, despite all 5kw being used (the other half would show up as unknown).

I just checked, and it’s an '04 model, with some oxidation going on at the input pipe. I’ll keep an eye on it, I feel like it’d be less of a headache to replace if it acts up. It’s just about the last appliance we’ve had to address since moving in; fridge, heat pump, and stove have all needed work in the past year.

Dont worry about your heater failing. Sense does that with mine too. I just figure that things will get better as it learns more about the heater.

Here’s what my heater looks like:

1 Like

Ditto. I’ll let it ride next time it’s found :slight_smile:

It is also possible the lower element is wired for 120vac via a return path through “ground”. There appear to be several ways to wire dual element hot water heaters.

Ahh… the joy of owning a home :money_mouth_face:

yikes!

I thought the same but a Google search turned up a couple of schematics
where the lower element was wired to a ground symbol.
I did see one person on this site observing his Sense was reporting hot
water with 120Vac sometimes.

Alan Kirk GMail

After a couple of weeks, the water heater was found again, but the problem persists. Seems to accurately ID it in the morning, and then in the evening it either misses it completely or only gets half the draw. When it comes on in the morning, there is no solar; in the evening the solar is just tapering off. Completely anecdotal, but curious.