IoTaWatt home energy monitor

I love data and I get why this is a draw for anyone who is data driven. However, 90% of most energy usage is going to be 4 sources:

1. Primary heating/cooling
2. Food cold storage
3. Hot water
4. An electric clothes dryer

There is just nothing in the same broad category as these items. And there is little you can do about it, other than be mindful of thermostat settings and, maybe, do something with your hot water when you're out of town.

It is a lot of effort to chase pennies with some of these other, mid-tier energy applications. This is like the folks who chase phantom power. Sure, it's a great idea in theory, but at the end of the day it likely isn't worth it to have to start everything up, every time. Again, might be something to think about for a long vacation, but otherwise it is not an effective use of time or resources.

quote:

---

but I think real-time energy monitoring can be very useful for automating things.

---

True, but I think direct application of "internet of things" is going to be more efficient and adaptable than using power as a proxy.

Again, my 2 cents and probably not worth that.

quote:

---

For the rest of them, I think using power as a proxy for status and health of the appliances is the simplest and cheapest way to get some insight. I don't think it will be difficult to set up some calculations for long term average on runtimes and wattage, and from that determine whether something might be failing or wearing out. For example, if your fridge that typically consumes 50 kWh/month starts creeping up, you can get it repaired or replaced before you waste a fridge full of food or unknowingly pay an extra $10/month on power for a year or two.

---

This would be my danger zone - once I start to gather data, where does it end? The possibilities are endless!

In your fairly narrowly drawn case (maybe not so narrowly, as the Boomers are rushing into old age and decrepitude), it is an intriguing and composite (if not quite comprehensive) solution.

quote:

---

I also think there's a safety aspect to monitoring the whole home.

---

I'll even step back - having to do something like this, remotely, you probably do have the best overall solution, however flawed some of the aspects might be. Data, analysis and control in 1 piece.

quote:

---

This would be my danger zone - once I start to gather data, where does it end? The possibilities are endless!

---

Yeah that's my "problem" too. I started this thread to gather more ideas.

quote:

---

In your fairly narrowly drawn case (maybe not so narrowly, as the Boomers are rushing into old age and decrepitude), it is an intriguing and composite (if not quite comprehensive) solution.

---

Yeah next I need to figure out how to monitor gas consumption since the stove and oven are gas. I don't expect or even care to ever recoup the costs in savings in this case. It's 100% about safety, and monitoring utilities seems to be a cheap, simple, and non-invasive way to achieve that.

quote:

---

I'll even step back - having to do something like this, remotely, you probably do have the best overall solution, however flawed some of the aspects might be. Data, analysis and control in 1 piece.

---

We can call it "remote" monitoring in the loosest sense, she only lives 2 blocks away.

I don't know if I can think of a better way to spend $300 for the peace of mind it will provide us.

Would you mind posting some pictures of how it is connected in the panel?

quote:

---

Would you mind posting some pictures of how it is connected in the panel?

---

Here is a photo showing how the CT's (current transformers) get installed inside the panel:




They are split core transformers, so you just pop them open and clamp them around the wires you want to measure. There are no electrical connections to be made inside the panel.

At the top are two 200 amp CT's on the mains, and there are eight 50 amp CT's on select circuits. The top left and bottom right circuits are 240v, so you just clamp over one of the legs and tick the checkbox in the IoTaWatt settings to double the power for those.

I will likely add more CTs later (I can add 4 more to this unit as it has 14 channels), but for the time being I have configured IoTaWatt to lump the un-measured circuits into a 'misc' figure by subtracting all the measured circuits from the total seen on the mains.

I am still working on automating the process of identifying various devices and hopefully being able to detect "leaky" outlets and switches and such, but for now IoTaWatt has some basic charting tools built in that I have been toying with. I've collected about a week's worth of data so far, and here is what that looks like:


Cumulative kWh:



You can see here that the AC circuit consumes almost exactly 50% of the house's total power use, but also the blower fan for the AC (light green) uses a big chunk itself putting the total for cooling the home at about 190 kWh in one week. Total power consumption for the week was about 310 kWh, which projects to ~1200 kWh for the month. This converts to an expected energy bill of ~$170, which is pretty typical for this home at this time of year.


Next is instantaneous total power for a period of 1 hour:



You can probably see why it might be difficult to decipher what is going on by only monitoring the mains. I think this is why the systems like Sense don't seem to do very well with their machine learning tools.

But if we look at the same chart with individual circuit monitoring:



Then break out the AC & blower:



We are left with a pretty clear picture of what's happening:



But I still have a lot of unknowns in the unmonitored circuits:



I have no idea what that 1,000 watt load is that ran for 15 minutes.


Anyway, this is the results so far of me playing with this new toy. There is plenty enough data here for me to set simple rules in Home Assistant to notify about loads greater than an arbitrary threshold on arbitrary circuits for longer than arbitrary lengths of time, which I think is good enough to detect hair irons, space heaters, or other heat-generating devices that have been left on and possibly forgotten. Over time I will either find some machine learning tools to identify more things, or I may develop my own models.

I think this is hella fun and it will become more useful over time.