Utility provider (So Cal Edison) estimated EV charging use way off

[grayguy]

Does anyone else get estimated monthly usage from their electricity utility each month showing how they compare to neighbors, with usage calculating estimates for EV charging? So Cal Edison sends me one every month, and they seem to be way off with what portion of my kWh usage is attributed to charging. I have two Teslas, am on their Time-Of-Use Plan, and only charge when TOU kicks in. You'd think they could easily figure this out with their smart meter as well.

Below is what I get from Edison, attributing only 86 kWh to EV charging. My Tesla app shows that I've used a combined 575 kWh between the two cars for the month of April. They use different days of the month for their start/stop days, but that estimate seems in line and should be about the same percentage of use.

 

[jjrandorin]

I have similar information on my SCE bill although I am not on a time of use plan. I dont pay any attention to them at all (like Zero), though since it doesnt matter to me in the slightest what SCE thinks "efficient neighbors" use vs what I use. I have a 3300 sq. ft. home and from near the end of march to about august, I either have a negative balance of kWh used, or a very small (<100kWh) positive balance in Sept / Oct. This still puts me slightly worse than "efficient homes" which I call BS on, since my electrical usage is less than what it would be to run a fridge all month, if I didnt have solar.

Since this is currently posted in the Tesla Energy subforum (which is for solar or solar + battery discussions), how much are you generating with your solar install to still have a 1144 kW usage number?

 

[LoudMusic]

I suspect they are using algorithms on your consumption to identify different devices switching on and off.

If you're drawing 3A and then it goes to 35A they would assume a 32A device turned on. Then if it goes to 33A a 2A device turned off and the 32A and a 1A device are still running. Once they determine some regularity of these things happening on your circuit they can start to identify that the same object is making each of these power draws. Then match up the profile of each identified object to a database of known devices.

Problem is, there's many devices that can be very similar. An EV that charges at 240v32a for an hour solid might look an awful lot like a cloths dryer, water heater, or even an air-conditioner. Electric stoves and ovens probably will switch on and off several times during a use session.

They might need to tweak their algorithms and/or profiles to catch stuff like the initial power ramp that EVs seem to make during AC charging.

 

[BGbreeder]

I suspect they are using algorithms on your consumption to identify different devices switching on and off.

If you're drawing 3A and then it goes to 35A they would assume a 32A device turned on. Then if it goes to 33A a 2A device turned off and the 32A and a 1A device are still running. Once they determine some regularity of these things happening on your circuit they can start to identify that the same object is making each of these power draws. Then match up the profile of each identified object to a database of known devices.

Problem is, there's many devices that can be very similar. An EV that charges at 240v32a for an hour solid might look an awful lot like a cloths dryer, water heater, or even an air-conditioner. Electric stoves and ovens probably will switch on and off several times during a use session.

They might need to tweak their algorithms and/or profiles to catch stuff like the initial power ramp that EVs seem to make during AC charging.

Unless you have a concrete piece of information on this one, I think that you might be attributing way too much intelligence to this particular utility. Way, way, too much. (Speaking as a customer with more than a little experience with them in both routine and emergency operations...)

 

[LoudMusic]

Unless you have a concrete piece of information on this one, I think that you might be attributing way too much intelligence to this particular utility. Way, way, too much. (Speaking as a customer with more than a little experience with them in both routine and emergency operations...)

That's how I understand this consumer product to work.

Sense.com – The Sense Home Energy Monitor

sense.com

This guy did a decent explanation of how it works, nearly as I can tell.

Seems like a utility meter with network connectivity could have that type of hardware built in for a tiny amount more cost to the provider, and give them the ability to offer incredible amounts of information to their customers, and be able to sell that information to data mining / statistics companies.

 

[grayguy]

I have similar information on my SCE bill although I am not on a time of use plan. I dont pay any attention to them at all (like Zero), though since it doesnt matter to me in the slightest what SCE thinks "efficient neighbors" use vs what I use. I have a 3300 sq. ft. home and from near the end of march to about august, I either have a negative balance of kWh used, or a very small (<100kWh) positive balance in Sept / Oct. This still puts me slightly worse than "efficient homes" which I call BS on, since my electrical usage is less than what it would be to run a fridge all month, if I didnt have solar.

Since this is currently posted in the Tesla Energy subforum (which is for solar or solar + battery discussions), how much are you generating with your solar install to still have a 1144 kW usage number?

I do not have solar on my house. I could not find a more appropriate category on the forum for this question, as it is not specifically related to one of the vehicle models and is an energy-related question.

I can see how you would want a "net" usage after power production number compared to your neighbors, but that is a different question than how much energy your home consumes compared to your neighbors. That number would essentially just be showing you if your solar contractor under or over-calculated the number of panels you need to offset your household consumption billing per whatever Net Energy Metering tier you're on.

 

[grayguy]

I suspect they are using algorithms on your consumption to identify different devices switching on and off.

If you're drawing 3A and then it goes to 35A they would assume a 32A device turned on. Then if it goes to 33A a 2A device turned off and the 32A and a 1A device are still running. Once they determine some regularity of these things happening on your circuit they can start to identify that the same object is making each of these power draws. Then match up the profile of each identified object to a database of known devices.

Problem is, there's many devices that can be very similar. An EV that charges at 240v32a for an hour solid might look an awful lot like a cloths dryer, water heater, or even an air-conditioner. Electric stoves and ovens probably will switch on and off several times during a use session.

They might need to tweak their algorithms and/or profiles to catch stuff like the initial power ramp that EVs seem to make during AC charging.

They must have really faulty algorithms. My older Tesla Wall Connector is on a 60A circuit, and it's the device that draws 48A in the house.

 

[BGbreeder]

That's how I understand this consumer product to work.

Sense.com – The Sense Home Energy Monitor

sense.com

This guy did a decent explanation of how it works, nearly as I can tell.

Seems like a utility meter with network connectivity could have that type of hardware built in for a tiny amount more cost to the provider, and give them the ability to offer incredible amounts of information to their customers, and be able to sell that information to data mining / statistics companies.

Sure, a sense like meter and AI could be built in. However, the utility on question didn't, and they don't digitize the load data at a high enough frequency to do it. Even if they did digitize at a high enough rate, their network doesn't have the bandwidth to support large data volumes, so they would need to put enough smarts in the meter to do the analysis, at which point it isn't cheap.

BTW, you might want to read up on the Sense product; while it sounded good in theory, in practice, many owners have found that even after a year it struggles to identify loads. (There are a number of threads here.) For anyone experienced in power and algorithms it had, and, I believe, still has real issues with what they are trying to do. Not every load will have a signature start/run/stop pattern that is unique, or reliably similar. I didn't and don't think that it is easy. Finding a 10kW charger starting is one thing, telling the living room lights from the dining room lights is much harder.

My point I was endeavoring to make was not whether load analysis could be done, just that PG&E doesn't do it. Their bill average math seems to take very generic data and project it on to consumer's particular bills, neglecting details such as microclimates, and whether, say, the EV in question goes to the local grocery store once a week, or is used daily to go 125 miles.

 

[wwu123]

You guys give way too much credit to the utility to suggest they even care to attempt to do what Sense has completely failed to do when their survival depends on it.

I think the only way SCE knows or cares you even have an EV and charge at home, is that somewhere along the way you did a "home energy checkup" type web survey, and answered you do, and they retained and used that cookie in the crudest possible way. Maybe taking all the survey responses and dividing total miles driven by all households who answered, whether they owned an EV or not. Why they would break out EV charging, and attempt to break out nothing else, also is curious.

As BGbreeder stated, PG&E's doesn't do any load analysis. Moreover, in my neighbor comparison with 100 nearby homes, they only compare similar homes based on floor size and type of winter heating. The "average" monthly kwh use is slightly negative during summer, so obviously there are many solar homes like mine. But they are pooling homes with and without solar, with and without EV's, in the comparison, so it's utterly meaningless to benchmark with their comparison of "similar" homes...