Accuracy of plug-in electric-energy meters

[Sprucebeach]

I don't yet have solar but my issue is of concern to anyone interested in measuring electric-energy use.

I'm charging my PHEV from a 120V wall socket using the car-manufacturer's charger. I started monitoring the energy required with a Poniie PN2500 meter, and I've since supplemented that with measurements from a Poniie PN2000 and a Kill-a-watt P4400 meter. In addition, the car's app reports the kilowatt-hours charged.

In both partial and full charging sessions over several months systematic differences have appeared in the energy reported by the various methods. Referenced to the killowatt-hours reported by the most-expensive method (not counting the car itself!), the PN2500, the results are:

The Poniie PN2000 averages 5.8% higher,

The Kill-a-watt P4400 averages 6.8% higher,

And the car's app averages 15.5% higher!

I'm not sure how to think about all this. First, if the car's app is correctly reporting the energy used then, at current gas prices and with our high summer electric rates, driving in hybrid mode and driving in home-recharged electric-only mode cost about the same. That's disappointing but I don't know if it's real.

It's unsettling enough that all three plug-in meters report energy consumption that's significantly (and consistently) below what the car reports, but it's also disappointing that there are significant and consistent differences among those meters, with the most-expensive one reading the lowest.

I don't have any other way to measure the energy required to recharge my PHEV. Does anyone have any suggestions or any comments on the behavior I've observed?

 

[Warpspeed]

All sorts of difficulties measuring ac power with extreme accuracy.
Your sine waves are often quite badly distorted, and that leads to measurement errors.
Also the voltage and current will not always be exactly in phase.

While a power meter may be correctly calibrated with perfect clean supply and perfect linear load, feeding it with distorted waveforms can cause different power meters to give different results. Less than ten percent difference is probably not an unreasonable expectation.

 

[wpns]

What are the published resolution and accuracy for each of those meters?
There are all kinds of options for measuring power consumption, the Emporia Vue seems to get good reviews.

 

[Sprucebeach]

The PN2000 and PN2500 meters claim Class 1 accuracy, which is "1% at full rated current", which is 15 and 25 amps respectively for the meters. (I charge at 9A.)
The P4400's accuracy spec is "+/- 0.2%".
The PN2000 and P4400 display KWh to two decimal places, the PN2500 to three. My charging sessions are in the range 15-30 kwh.

 

[MattiFin]

Car reading is possibly energy input to battery. In addition you get couple of percent losses from AC-DC onboard charger.

 

[Sprucebeach]

But that's in the wrong direction. The car is reporting receiving more energy than the three meters say is being pulled out of the outlet.

 

[ThaiTaffy]

Check the power factor reading on the meters then work out apparent power from that and see if the cars reading is closer.
When measuring DC power we just use current X volts but AC meters will typically measure (current X volts) x power factor, this possibly could be the cause of your differing readings atleast between the AC meters and the EV.

 

[Sprucebeach]

Check the power factor reading on the meters then work out apparent power from that and see if the cars reading is closer.
When measuring DC power we just use current X volts but AC meters will typically measure (current X volts) x power factor, this possibly could be the cause of your differing readings atleast between the AC meters and the EV.

Thanks for your reply. Unfortunately, I didn't note the PF in any of my charging sessions. I'll be charging it again on 8/7 and will check this out.

 

[wattmatters]

It is unusual to see AC energy measurements at point of supply being lower than what the car reports as being delivered to the battery. The car has energy losses in the AC to DC conversion, and it may also have other parasitic losses, e.g. my EV will also charge the auxiliary 12V battery as necessary during a charging session and it is powering the car's management systems as well.

I would test all your plug in monitors with another load to see what discrepancies arise. Perhaps start with a pure resistive load (plug in resistive element heater) to eliminate power factor, then move to a more inductive load such as a pump.

 

[Warpspeed]

Thanks for your reply. Unfortunately, I didn't note the PF in any of my charging sessions. I'll be charging it again on 8/7 and will check this out.

Power factor is one possible source of error, as only real true watts are usually measured, not reactive volt amps.
Another problem is harmonic distortion, which can be different on the voltage and current waveforms and constantly changing.

Its rather like trying to measure the precise dimensions of a potato with extremely accurate calipers.
What you are measuring is extremely irregular, and precise measurement difficult and unrepeatable.

Just because the calipers are guaranteed accurate, does not always mean the measurement of something irregular is meaningful.

 

[wattmatters]

Yesterday my Iotawatt reported 13.34 kWh delivered to the garage's sub-board for the whole day. The EVSE has a small draw of ~1.8 W when idle so ~40 Wh for the time not charging yesterday. So call it 13.30 kWh

My EV charge service reported 12.967 kWh (it keeps tabs on what the EVSE itself reports).

My car's charge status reported increasing from 33.5 kWh to 46.1 kWh, so +12.6 kWh. The car's on board data is only sampled every so often, so is subject to some error.

Iotawatt at MCB -> EVSE (garage) -> car's DC reporting

13.30 kWh -> 12.967 kWh -> 12.6 kWh

 

[Warpspeed]

13.30 kWh -> 12.967 kWh -> 12.6 kWh

That is pretty good, and about what might be expected.

 

[wattmatters]

That is pretty good, and about what might be expected.

And to illustrate my point about auxiliary battery using some of the power while charging the car, can see the 12 V battery voltage hovers between 14.2 and 14.3 V during the EV charging session:

 

[Sprucebeach]

Check the power factor reading on the meters then work out apparent power from that and see if the cars reading is closer.
When measuring DC power we just use current X volts but AC meters will typically measure (current X volts) x power factor, this possibly could be the cause of your differing readings atleast between the AC meters and the EV.

Now I've had the opportunity to look at this. Over a full charge cycle the power factor hovered at about 98.9% - far too high to explain the meters reading 10-15% lower than what the car reports.

Incidentally, the discrepancy between the meters and the car is pretty linear with all-electric miles traveled. This means that the car isn't just adding some hypothetical number of kilowatt-hours at the end of every charging session to what the car is actually accepting. Instead it's reporting proportionally higher values than whatever the meters are reporting.

 

[Sprucebeach]

Yesterday my Iotawatt reported 13.34 kWh delivered to the garage's sub-board for the whole day. The EVSE has a small draw of ~1.8 W when idle so ~40 Wh for the time not charging yesterday. So call it 13.30 kWh

My EV charge service reported 12.967 kWh (it keeps tabs on what the EVSE itself reports).

My car's charge status reported increasing from 33.5 kWh to 46.1 kWh, so +12.6 kWh. The car's on board data is only sampled every so often, so is subject to some error.

Iotawatt at MCB -> EVSE (garage) -> car's DC reporting

13.30 kWh -> 12.967 kWh -> 12.6 kWh

Yes, those numbers make sense - they're both *lower* than what your lottawatt says the garage board is delivering. My problem is the opposite - the car is consistently reporting *higher* numbers than what my three meters say is coming out of the wall.

 

[wattmatters]

Yes, those numbers make sense - they're both *lower* than what your lottawatt says the garage board is delivering. My problem is the opposite - the car is consistently reporting *higher* numbers than what my three meters say is coming out of the wall.

Yes, hence why I said:

It is unusual to see AC energy measurements at point of supply being lower than what the car reports as being delivered to the battery.

I don't have an explanation other than the car is not accurately measuring/reporting battery power and energy.

What sort of PHEV is it?

 

[zanydroid]

+1 to "Which car is it?" I trust external energy meter a lot more than the car. Have you checked the forum for the car? It may be a known issue.

There are a decent number of places where grid charging even off-peak is comparable to running the car off gas. California for instance.

Charging at 120v*9A is probably very inefficient, since a lot of cars don't do much to scale down their fixed overheads when charging that slowly.

TBH I would put all of those plug-in meters in the same quality bucket. <$40, right? And +/-5% is sort of the margin I expect with my on-grid energy monitoring.

 

[Sprucebeach]

Yes, hence why I said:


I don't have an explanation other than the car is not accurately measuring/reporting battery power and energy.

What sort of PHEV is it?

Sorry, wattmatters. I missed that you were the one who had made that comment.

The car is a BMW X5 50e. I think your suggestion that the car isn't accurately reporting the kWh required to recharge the battery is correct. Unless there's some systematic error in these three $30-60 plug-in meters, it's unlikely that all three would be inaccurate in the same direction and by about the same amount.

I find the BMW app to be suspect anyway. If you fully charge the car and drive it around in electric mode the electric "mileage" the app reports is preposterously better than what you calculate if you measure the energy you have to put into the car to bring it back to full charge, which is really the only electric mileage that you care about.

In my experience, in electric mode I can go about 1.7 miles per kWh I put into the car. The app typically reports numbers like 2.3 miles/kWh. And if I use the app's own reported energy required to recharge - always higher than what I measure - my mileage is about 1.5 miles/kWh.

In hybrid mode, never plugging the car in, I get about 24.5 mpg with mid-grade gas and the kind of driving I generally do. As I mentioned, with my area's current gas prices and summer electric rates the two driving modes cost about the same if I believe the app's reported recharge energy.

Thanks for your thoughtful comments.

 

[zanydroid]

2.3 is much more in line with the gas mileage and my expectations of the vehicle class, than 1.5

1.5 is more like a full size SUV and approaching truck

 

[Sprucebeach]

2.3 is much more in line with the gas mileage and my expectations of the vehicle class, than 1.5

1.5 is more like a full size SUV and approaching truck

I agree. It's too bad that neither my nor BMW's measurements can confirm that 2.3. I'm guessing that BMW is reporting miles/kWh based on the amount of energy the electric motor is consuming. Unfortunately the amount of energy the driver has to put back into the car includes the amount required for the electronics, the climate control, sound system, wipers, lighting, etc. plus energy lost to inefficiencies of the charging process (as you pointed out). This is the actual cost to the driver of electric driving and what yields the 1.5 or 1.7 miles/kWh, depending on which measurement of charging energy you believe.

As I was writing this I remembered that for two months earlier this year I was in a location where I *was* able to do 240V charging. I don't have my own measurements of the energy charged but I have the app's value, and the app also shows the odometer readings when charging was done. The trips were shorter but I had 12 of them from which I could extract the electric miles traveled and energy recharged. The average of those is 1.68 miles/kWh, compared to 1.48 for 120V charging. That would seem to confirm your belief about efficiency. However, that conclusion is blurred by the 240V location being flat terrain with little high-speed driving while my home 120V location has lots of hills and lots more high-speed driving. I guess the only way to know for sure would be if I installed a 240V outlet, and there doesn't seem to be an economic justification for that.

 

[wattmatters]

In my experience, in electric mode I can go about 1.7 miles per kWh I put into the car.

That about double the energy consumption of my MG4 EV.

In hybrid mode, never plugging the car in, I get about 24.5 mpg with mid-grade gas

And that's about 60% higher fuel consumption than the Mercedes GLC250d I had before getting the MG4.

That's a thirsty car you have!

Unfortunately the amount of energy the driver has to put back into the car includes the amount required for the electronics, the climate control, sound system, wipers, lighting, etc. plus energy lost to inefficiencies of the charging process (as you pointed out). This is the actual cost to the driver of electric driving and what yields the 1.5 or 1.7 miles/kWh, depending on which measurement of charging energy you believe.

If they are not reporting the power drain from the battery, then that's would be pretty misleading. I'd expect the rest of it to be something in the order of 10-20% extra maybe, depending on the type of cooling/heating system used, and overheads for battery thermal management.

Yours seems excessive.

I guess chatting with others on a BMW owners forum who have the same car could at least validate if this is a common issue with this car model or something unique to your car.

 

[zanydroid]

This is the actual cost to the driver of electric driving and what yields the 1.5 or 1.7 miles/kWh, depending on which measurement of charging energy you believe.

I was referring to the Wall to wheels number including all overhead. On my Ioniq 5 it's 3-3.5, on a Rav4 PHEV it's 2.5-3.

1.5 feels like a resistive battery preconditioning or resistive heaters mildly on. The only time I see 1.5 on my HI5 is when the PTC heater loop is prepping the battery for DCFC.

+1 to BMW forum for this car. That will 100% sort this out, both from empirical numbers and from telling you if buying a OBD scanner and recording relevant timeseries during a drive, will give you the ground truth. It is also generally tested by review websites. You can take the accessible battery kWh (probably need to get it from forum to de-risk the manufacturer potentially reporting full battery capacity including headroom) and divide by the range numbers from the reviewer test.

EPA has the numbers too for a previous gen of the PHEV... and actually it might truly be 1.6 mile/kWh. LOL. https://www.fueleconomy.gov/feg/Find.do?action=sbs&id=45119