Discrepancy between grid usage reported by Tesla app and by electricity bills

Discrepancy between grid usage reported by Tesla app and by electricity bills

I've done some analysis of the data reported by the Tesla app vs from my electricity bills, and I've found that my electricity company thinks I'm using about 0.4 to 0.5 kWh more each day than the Tesla app reports. In fact on days when I'm theoretically completely self sufficient (all power usage from solar and battery) I'm always billed a minimum of 0.5 kWh per day. On such days the Tesla app usually only reports 0.1 kWh of grid usage.

Looking at the hour-by-hour breakdown from my electricity retailer, it appears that a small amount of grid power is being used whenever the battery is charging or discharging, but not when it is idle (when the battery is fully charged and excess solar is being exported back to the grid). This seems to imply that the Powerwall is using some grid power that is not reported to the gateway. Does anyone know if this is expected?

I know it's not a huge difference but I'd like to understand what's going on.

sashton | 18 februari 2018

From my experience the Tesla sensor is not particularly accurate. I don't know whether its sampling rate is low or it is simply the standard it is built to. Bizarrely I have three import/export meters in series here. The one provided by the utility company and the Fronius one agree with each other to within fairly small tolerances. Both are certified to EN50470 class B which should ensure accuracy to within 1%. The Neurio sensor that the Tesla uses claims the same accuracy (1%) but, as far as I can find, only appears to have certification for safety.
Regarding your second point about a small amount of grid energy being consumed while charging and discharging. I take 30 second samples from the Fronius meter and apart from the times when the battery is coming out of standby I do see small amounts occasionally flowing in or out but rarely more than 20 watts. This situation was considerably worse until I switched from a WiFi connection to my grid sensor to an RS485 wired connection.
As you mention, since this is close to the tolerances of the claimed sensors and considering there must be some reaction delay in the system, the odd fraction of a kWh/day doesn't seem to bad a price to pay.

Tesla-David | 18 februari 2018

I would agree with @sashton, in WA state we pay $0.09/kWh, and that would amount to 4- 5 cents/day nothing to get overly concerned about IMHO.

shadaway | 9 augusti 2018

Follow up question. What would you say if the KWH for the month on the powerwall2 were 20 and the meter is reading 203. This is a 10x difference. I've been searching on this topic for awhile now and this is the first thread I've found.

Thank you in advance.

cwied | 11 augusti 2018

If you have a large discrepancy between what the Powerwall app reports for your site usage and what your meter is reading, the most likely explanation is there is a problem with the way the Current Transformer (CT) for the Powerwall is installed. It might be installed on the wrong wire, or maybe the CT itself is faulty. Call Powerwall support and report the issue to them.

jm1991 | 29 augusti 2018

Just installed Powerwall 2 at our home. I think we have a CT problem. On the app with the four quadrants (grid, powerwall, solar, and home consumption), at the initial install, as soon as the Powerwall was placed in discharge mode, the CTs reported a negative number for home consumption---even though home consumption was running at 7Kw just before. They did something with the CTs (moved them? reversed a polarity?), and then consumption reported a positive number. This fix cause yet another strange reading---one that they haven't figured out how to fix: now, if the home consumption is reporting, for example, 7.5Kw, the moment the Powerwall goes into discharge (and is feeding into the home), the home consumption jumps up by 5Kw to 12.5Kw and the Powerwall indicator shows 5Kw being discharged into the house. This, of course is backwards, as the home consumption should drop by 5Kw to a read of 2.5Kw instead of adding the Powerwall's power. I think I'm on the right track as to the symptoms, but I do not know what to do to fix this. Tesla acknowledges that this is a weird reading, but they haven't figured out a fix yet. Has anyone else had this type of app-monitoring-inaccurate-reading issue?

cwied | 29 augusti 2018

Actually, the home consumption shouldn't change based on Powerwall discharge. I haven't had this symptom myself, but from your description, one possibility could be that the CT is measuring home load but is configured to say it is measuring the site current. If the draw from the grid remains 7.5 kW and the Powerwalls are outputting 5 kW, that would mean home load would be calculated as 12.5 kW.

You could try running through the setup wizard to see what the CT page shows when your Powerwalls are discharging. If you see 7.5 kW on "site," you could change that to "load" (or something like that - not sure of the exact term) instead to see if it fixes your issue.

dmunz | 21 september 2018

I'm new to the Powerwall 2 but I've had solar for a couple of years now so I'm trying to align all the different reports.

For solar production, the app runs about 98% of the daily number reported by my SolarEdge which is reasonable.

But the consumption and grid numbers are nuts.

For Home Usage the daily screen consistently shows flat a zero except between around 8AM and 4PM. There is no way we are not using power during the times it reports zero.

The grid display also shows use during this time but is negative (to grid) the rest of the time.

It looks very much like the system is reporting inverted. That is, the times when I'm putting power to the grid (daylight hours) are showing as pulling from the grid and at night I'm shown as pushing out to the grid.

Is that a possible error and how can I correct it?


dmunz | 28 september 2018


I now have a full weeks data from the App and my smart meter and the flows are clearly inverted. The data from the app is 180 degrees out of sync with the data from the smart meter.

Is this something I can fix?

It's not such a big deal for Grid use as I can invert the data myself but it makes me question the reporting for Home Usage and Battery. (Solar is good as I said above.)

TIA for any advice.

sashton | 28 september 2018

If you are suggesting one of your sensors is misreading flow and your "180 degrees out" means it is registering import as export then all you have to do is unclip that CT clamp and connect it to the cable the other way round.
Failing that you can swap the CT cables at the Neurio end.
It does seem to be such a fundamental error that it seems surprising your installer would have not seen the issue as the wizard displays the readings in real time as the sensors are configured.

NB Of course, do take care if you plan the adjust the CT clamp and observe suitable electrical safety measures.

dmunz | 29 september 2018

Sashton, Thanks for that. I'm not actually that familiar with the internals so I don't know where the CT is (or what it even looks like!) But I will try to sort that out.

Just to show the data, in addition to the Tesla app and smart meter data, I have individual sensors on my circuits (these are Panoramic sensors which is a part of Centrica who I work for.) Copying this in will be messy using text, but this is what I am trying to sort out. Where I say FROM I mean above the line in the app and TO is below the line.

SE – Solar Edge Reported Solar Production
TS – Tesla App Reported Solar Production
FPW – Tesla App Reported FROM Powerwall
TPW – Tesla App Reported TO Powerwall
PFG – Panoramic Sensor Reported FROM the Grid
SMF – Smart Meter Reported FROM the Grid
TTG – Tesla App Reported TO the Grid (below the line)
TFG – Tesla App Reported FROM the Grid (above the line)
SMT – Smart Meter Reported TO the Grid
THT – Tesla App Home Use Total
PPW – Panoramic Reported Across the Powerwall Breaker (the is bidirectional so it shows both to and from the Powerwall which lines up for the most part with FWP+TPW)

9/17/18 37.8 | 36.9 | 98% | 14.8 | 15.2 | 43.9 | 43.2 | 43.0 | 27.5 | 27.5 | 44.5 | 30.22
9/18/18 40.0 | 39.0 | 97% | 13.1 | 15.3 | 42.6 | 42.9 | 42.7 | 24.4 | 24.4 | 28.3 | 28.74
9/19/18 39.5 | 38.5 | 98% | 13.0 | 16.2 | 46.7 | 46.7 | 46.5 | 26.7 | 26.7 | 42.9 | 29.22
9/20/18 21.8 | 21.2 | 97% | 14.3 | 10.0 | 36.0 | 38.7 | 38.4 | 21.0 | 21.0 | 24.4 | 24.66
9/21/18 20.7 | 20.1 | 97% | 09.2 | 09.1 | 32.1 | 38.9 | 38.6 | 15.3 | 15.3 | 20.8 | 19.75
9/22/18 12.6 | 12.2 | 97% | 09.0 | 08.4 | 24.9 | 31.1 | 30.7 | 07.9 | 07.9 | 09.4 | 18.91
9/23/18 37.2 | 36.3 | 98% | 09.9 | 16.7 | 38.5 | 36.8 | 36.6 | 22.3 | 22.3 | 32.5 | 26.84

Of course this is really a limited data set but interesting (to me) anyway.

One other issue I have with the Tesla app is that it shows continuous production FROM solar. This is very comb-like with the spike aligning with the AC unit running during dark hours.

I guess, the whole point for me is can I trust the data from the app and what can I do to clean it up? It would also be really nice to get data directly over longer time periods. For now, I just manually copying what I read on the app each day into a spreadsheet.


dmunz | 29 september 2018

Well, that didn't work out to good...

the columns should line up with no heading for the dates or the % data (this is difference between the Solar Edge and Tesla solar reporting.)



sashton | 29 september 2018

Judging by what you have there I would say your solar sensor appears to be working correctly. Your daily solar production reported by Tesla seems fairly close to your daily solar yield reported by Solar Edge.
It is difficult to make sense of your grid readings as on the surface they appear to disagree dramatically with each other. It would be easier to work out the issue with more granular data.
I have an intermittent problem with my Tesla solar sensor. Occasionally it loses contact with the gateway. The net result is that until it is resolved by restarting the gateway the PW stubbornly stays in standby and reports that the solar gain is 0 but the house itself is generating as shown by the black line here:
Anyway it doesn't look like your CT clamps are misaligned but fyi these are what they look like (assuming in your geo Tesla use the Neurio sensors) :

lizfg63 | 7 april 2019

I’ve just received my first power bill since the Powerwall 2 Installation and it tells me I’ve tripled my energy consumption! Tesla app tells me I’m 97% self powered. Anyone have any ideas about what’s going on here? I haven’t changed any usage habits, if anything should be using less as it’s cooled down and the Air con hasn’t been on. I was without Solar for 1 month and the bill wasn’t even that high then. Slightly panicked by this.

roger.lever | 24 februari 2020

I am in a similar position to lizfg63, in that, having had solar panels for a few years, I have recently had smart meters installed and also Powerwall 2; at the same time I have changed my energy supplier. I live in the UK. Basically, there is a huge discrepancy between my consumption from the grid as indicated by the IHD for my smart meter and the TESLA Gateway. Today, so far, the IHD indicates a consumption of 31 kWh; whereas the Gateway, via the App, is indicating just over 18 kWh. I have not changed my energy usage throughout this period. The Gateway indicates solar flow accurately. The Gateway consumption ties in with my historic usage but I understand that smart meters ‘cannot lie’! I have contacted my installer and energy supplier but, as yet, have had no reply from either.

Noam | 24 februari 2020

I'd take a second look at the dates covered by those utility bills, and the corresponding dates reported by Tesla. It's entirely possible, especially if you've just switched on your system (or just got a new meter), that the utility billing cycle is for a longer duration for that first bill. It's also possible (in Roger's case) that the Gateway *wasn't* reporting all of his consumption (does your entire house run through the Powerwall, or is it only backing up a selection of circuits), and so his historical data is wrong too. But that's a less likely scenario. I'd look at the billing dates first. I'd also look at what the utility is reporting. Are they listing *all* of the power that flowed in to your home through the meter (and separately listing the power that flowed back out), or are they reporting the "net" difference between the two. Since your utility (in most cases) can't see the power that's generated from the panels and used immediately by your home, their consumption numbers generally won't match what is reported by Tesla - you need to know which numbers to add to get the corresponding values.

Bill Arnett | 4 juli 2020

I’m having a similar problem. I’ve had my Powerwall for several months (and PV panels for years). Until yesterday the readings in the app were totally wacko. But a Tesla tech visited me yesterday and now everything looks fine and consistent in the app and consistent with what my inverters report. BUT the numbers don’t match what my utility’s smartmeter says. It’s as if the meter is consistently (for energy flowing in both directions) reading lower numbers by about 50%. For example, my meter read 65 at midnight, it reads 55 now, so I’ve sent a net total 10 kWh back to the grid today. But the Tesla app says I’ve pulled 6.2 from the grid and sent 21.7 back for a net of 15.5 kWh. Where’s the other 5.5 kWh??? | 4 juli 2020

@Bill Arnett - There could be something wrong, but in some installations, there are loads outside of the PowerWall power. For example, you might have your HVAC with a breaker on the meter side rather than the PowerWall side. It would mean in a power failure, you will not get HVAC. It also means Tesla does not see that load, only the loads connected to the PowerWall side.

Another thing you might do is when the PowerWalls are at 100% and nothing is turned on in the house, look at the output of the solar inverters if you have a model with a display, not the app. Then look at the meter and see if the numbers agree. Then look at the app and see if it agrees. Hopefuly they should all agree.

Bill Arnett | 4 juli 2020

I do have a couple of loads that are not backed up: my two Tesla charging circuits. I do not want the cars to charge during an outage, saving the Powerwall’s energy for my house. But for this experiment I have unplugged the cars so that’s not an issue. (And anyway, the powerwall gateway and sensors are configured to add the unbacked circuits to the rest of the house load; I have verified that this works correctly.)

I have verified that the Tesla app is reporting the solar power correctly. My inverters report power and energy and the numbers match the app closely.

My utility meter also reports instantaneous power in addition to total energy. For example, this afternoon the Tesla app claimed that my system was sending 3.9 kW to the grid but the meter showed only 2.77; another time Tesla said 2.1 while the meter said 1.3. It’s as if there’s a large extra load Tesla doesn’t know about.

But the error is in the opposite direction after the sun goes down (Moon phase is irrelevant :-). I don’t have the exact numbers yet for today but my recollection from yesterday is that while Tesla said my house was drawing about 1 kW from the grid the meter said only 700 watts. As if there was an unknown generator in my system. I’ll have exact numbers in a few hours.

Throughout all this my Powerwall is at 100% and no power is flowing to it according to the app. I suppose I could turn off it’s breaker and see if that changes anything.

Bill Arnett | 4 juli 2020

OK, now I have data from the evening when both the solar and Powerwall are at zero. In the first case, the utility meter was reading 1.29 kW from the grid to my house while Tesla’s app was saying 2.2 kW. Another case had PG&E saying 620 watts from the grid while Tesla said 800. As if there was an extra bit of load that PG&E didn’t know about, exactly opposite from the situation earlier in the day.

In terms of energy Tesla thinks I have contributed 19.5 kWh net to the grid today but PG&E is seeing only 12.

It’s hard for me to believe that PG&E’s meter is broken. I suppose that’s possible but it seems much more likely that the Powerwall Gateway’s sensors are misconfigured. | 5 juli 2020

@Bill - Yep, I'm now sure one of the current transformers is misconfigured and/or flipped. I actually have the identical Neurio power monitor that Tesla uses in the gateway. I had it on my house before the Solar roof/Powerwalls. Setup is a bit complex and easy to get one of the current transformers backwards and/or one of the CTs installed backwards, which would cause the problems you're seeing. While I can get into the Neurio I installed, I cannot get into the Tesla one. These are configured remotely via the Web, but during the solar/PowerWall install, the installer did it from his phone.

The best option - Have Tesla fix it.

If you are a DIYer and are very comfortable working with 240VAC power, you could open up the gateway and/or meter panel and look for the current transfomers (CTs). One is likely backward or misplaced, although they are not marked really well. You could try and remove it and swap the direction. Two of them are part of the switchover and are not removable. This guide may give you some pointers and show what a CT looks like:

Note that if you got into the settings you can invert a current transformer in the software, but I don't think you can get access. Tesla could.

Bill Arnett | 5 juli 2020

OK, now I’m even more confused. It sure looks to me like Tesla is right and PG&E’s meter is wrong:

This evening when my solar is off and very little else is going on in my house I looked at the power (kW) indication on my SmartMeter. It read 0.8 kW. That seems reasonable. So I then turned on a hair dryer which I had previously measured as drawing 1.5 kW. Tesla reported an increase of 1.5 kW as expected. But the meter said only 1.5 kW which is an increase of only 700 W. WTF? Then I tried turned off the hair dryer and started charging my wife’s Model 3 at 8 kW. Again the Tesla app reported an increase of 8kW but the meter showed only and increase of 6 kW.

Is PG&E’s meter broken?

(During this the meter a couple of times briefly showed a code “ADL”. But I couldn’t find anything on the Web to show what that means.) | 6 juli 2020

@Bill - Hard to imagine the meter is not working properly. Perhaps it is very slow to respond on the display, but internally it is collecting the right data! I agree it does sound like the Tesla data is accurate.