Testing Smart Meter accuracy

[K8MHZ]

My POCO is switching out to GE i210+C Smart Meters. I got mine the day before yesterday.

Yesterday I did a quick read and just got the feeling it was high. I keep pretty good tabs on my electric use. With the old mechanical meters I could time the rotation of the disc with a stop watch, do some math and get an 'instantaneous' value. I did that a few times, comparing the meter's value to a Kill-a-Watt connected to the only load on at the time and they were always identical.

Now, with no disc to time and no other value except accumulated KWh on the display, it's going to take more than a stopwatch and a calculator to determine accuracy.

Here is my dilema. As a learned electrician that has kept up on smart meters, I would consider such an observation (a gut reaction that the meter is reading high) from a customer to be a result of hype and not necessarily true. But also as a customer that has kept track of electrical use for over a year, I am now beginning to wonder. If I could do a quick test, which I no longer am able to, my fears could be laid to rest. This also means I cannot do a quick test for a customer at this point in time.

So, an Internet search led me to this product:

http://www.optimumstores.com/ted-5000-home-electricty-monitor-kits.html

Even with the above, since the Smart Meters only have a resolution of 1 KWh, a test would have to take a couple hours minimum to make a valid comparison.

So, back to my situation.

The last two October billing cycles (2011 and 2012) I used 310 and 336, respectively. Last December's billing was 389. So, let's take the 389 and divide by 30. I get 12.96, or let's say 13, per day.

Now, it's been less than 48 hours and my meter already reads 30, or 15 per day, and I am still short more than an hour to hit 48 hours.

What I am seeing is what any customer can see with no special equipment. The increase of 2KWh per day will certainly raise the price of the bill. Also, being 2 off out of 13 is a substantial percentage and worthy of further investigation.

For now, I am going to keep a log, as the meter is very easy to read. I have the usage history for a year on the Internet.

I also have a Kill-a-Watt to measure individual loads, just in case the meter is correct and something is using more energy than before. Right now it's on the fridge. It's been 18 hrs, and racked up 1.63 KWh. So there is 2.17. I still have over 12 to locate, and the fridge is the biggest load......except for maybe the computer?

The gist of the matter is, if I can't answer these questions for my own situation, how am I supposed to do it for a customer?

If all of a sudden their use increases by 15 percent and the only thing that has been changed is the meter, don't you think they will be looking for some answers?

 

[zbang]

(My poco saves me a little time by putting the annual history on the bill, in addition to the calculated KWH/day. They also answer the phone .)

I think this may come down to a cost v cost problem- cost of the power v cost to measure. If you were to put a logger on the customer's house for, say, a week, what would you charge? If this is more than maybe a years worth of extra electric charges, how many customers would bother? My average is about 12 KWH/day @ about 14.4 cents/KWH. Another 2 KWH per day probably isn't worth my making a phone call. Might be for someone else.

There may be some money to be made by selling a package of a week or two's usage monitoring and a report. The report would simply be a description of the monitoring and methods, and a sheet/graph showing the usage over time and the totals. Once it's written, the report shouldn't take more than maybe an hour to produce and mail.

In the end, though, I suspect that it's more likely to show that the mechanical meter was running slow than that the electronic one is running 'fast'.

 

[K8MHZ]

(My poco saves me a little time by putting the annual history on the bill, in addition to the calculated KWH/day. They also answer the phone .)

I think this may come down to a cost v cost problem- cost of the power v cost to measure. If you were to put a logger on the customer's house for, say, a week, what would you charge? If this is more than maybe a years worth of extra electric charges, how many customers would bother? My average is about 12 KWH/day @ about 14.4 cents/KWH. Another 2 KWH per day probably isn't worth my making a phone call. Might be for someone else.

There may be some money to be made by selling a package of a week or two's usage monitoring and a report. The report would simply be a description of the monitoring and methods, and a sheet/graph showing the usage over time and the totals. Once it's written, the report shouldn't take more than maybe an hour to produce and mail.

In the end, though, I suspect that it's more likely to show that the mechanical meter was running slow than that the electronic one is running 'fast'.

I suspect you are correct. It would just be nice if they would have taken the customer a bit more into consideration and had the total watts being used as part of the display.

 

[K8MHZ]

We are supposed to get some sort of Internet info from the meter in January, but even the POCO isn't sure what that info will be.

I don't think the meter is capable of showing any instantaneous value. I think it just sends the accumulated KWh when queried. Also, the Internet may have the same 1 KWh resolution as the meter, along with a communications lag.

 

[mivey]

My POCO is switching out to GE i210+C Smart Meters ... With the old mechanical meters I could time the rotation of the disc with a stop watch, do some math and get an 'instantaneous' value...Now, with no disc to time and no other value except accumulated KWh on the display, it's going to take more than a stopwatch and a calculator to determine accuracy.

These meters still have that capability. In your case, the meter is a ten-segment emulated disk with three of the segments in black. If you look at the meter face, just under the 4 or 5 digit reading on the right side, there is a 3-segment display to the left of the "kWh" indicator.

With only 3 of the ten segments displayed, you will have 5 cases where the display is blank. You will see one or more black segments for the other five states. As the black segments roll in from the left you will go through 1->2->3 black segments. Then as they roll out on the right you will have 2->1 then back to blanks. If we use 0's and 1's you will go through the ten states (you can only see the first three digits) and have:
0001110000
0000111000
0000011100
0000001110
0000000111
1000000011
1100000001
1110000000
0111000000
0011100000

You will notice you see blanks (0's) for the first five states. Each of the ten state represents a Kt pulse. Each rotation (10 full states) represent a Kh pulse. The values for Kt & Kh should be on the face of the meter. For this meter type with a CL 200 it should be Kh = 10 watthours and Kt=1.0 watthours.

The disk simulator displays for other meters are not all the same as this but the idea is the same. This particular meter also pulses a 25 ms IR ouput with each Kt pulse and you can pick it up with a sensor over the LED (which we use when making meter tests).

 

[mivey]

I suspect you are correct. It would just be nice if they would have taken the customer a bit more into consideration and had the total watts being used as part of the display.

This meter does have the capabilty of displaying instantaneous kW updated every second but it is an option not normally enabled.

 

[gar]

121110-1208 EDT

K8MHZ:

For my meter here is the procedure to obtain relatively fine information on usage of energy:
http://www.griffmonster.com/miscellaneous/electricmeter/electric_digital_meter_reading.html

On the TED system. It is nowhere accurate enough to use as an adequate test of your smart meter. TED is a useful tool, but has many problems. Inaccurate on non-unity power factor, data communication problems, and changes or loss of historical data. Apparently big problems resulting from storm interruption of power. In other words some users have experienced huge changes in cumulative energy use. For certain purposes TED can be very useful, but not to check the accuracy of a smart meter.

The various graphs I have referenced from time to time are obtained from a TED 1000 system. See some of these plots at http://beta-a2.com/energy.html . With the 1000 System I get 1 second resolution.

Problems with my smart meter, but not the meter itself. I get various stories from the ignorant people at DTE when I ask questions about the smart meter system. The only people I have contact with are the ones via the main DTE phone number.

One of the arguments to the MPSC (Michigan Public Service Commission) in the DTE filing is that the customer via the Internet will be able to track their energy consumption in moderately fine detail. As far as I can tell this is quantized to 1 hour but with a 1 day delay. This seems to mean the smart meter is storing the average energy use for each hour. This should be distinguished from whatever cumulative energy register the meter may have. Various stories are given as to when this data is read and available. The most ignorant responses are only monthly data. That I already get on my monthly bill, and is already provided on the Internet site..

The most plausible information is that under normal conditions the meter is read between 12 midnight and 4 AM, and the previous day's data is transferred. Thus, in the morning I should be able to see the previous day's data.

How long does RF radiation occur from the meter. No useful data from DTE. Could be as short as a few seconds per day, or much longer if a mesh network is being used. I have not set up a monitor for this yet.

Technically as soon as my meter was installed and 1 day had passed I should have been able to get yesterday's data. But that is not so. Then someone told me that it took 30 days before I could get this information. Still not so. Now I have passed 1 full billing cycle. Still not so. Called again and got another ignorant person that is supposed to be knowledgeable on the Internet site. I know where the information is supposed to display, and this most recent person did not even know where that spot on their very poor web site was located. Then I was was told that they have had some customers that have not been able to get data. Inconsistent information from this person. She knew more than she was telling. So the buck is passed to the IT people. That was also done a week earlier and nobody responded to me. Namely their people do not know what they are doing or are trying to cover up problems they have.

I don't really want to continuously monitor my system, but I do want to understand how the system works so that I can accurately describe it to whoever is interested.

.

 

[K8MHZ]

These meters still have that capability. In your case, the meter is a ten-segment emulated disk with three of the segments in black. If you look at the meter face, just under the 4 or 5 digit reading on the right side, there is a 3-segment display to the left of the "kWh" indicator.

With only 3 of the ten segments displayed, you will have 5 cases where the display is blank. You will see one or more black segments for the other five states. As the black segments roll in from the left you will go through 1->2->3 black segments. Then as they roll out on the right you will have 2->1 then back to blanks. If we use 0's and 1's you will go through the ten states (you can only see the first three digits) and have:
0001110000
0000111000
0000011100
0000001110
0000000111
1000000011
1100000001
1110000000
0111000000
0011100000

You will notice you see blanks (0's) for the first five states. Each of the ten state represents a Kt pulse. Each rotation (10 full states) represent a Kh pulse. The values for Kt & Kh should be on the face of the meter. For this meter type with a CL 200 it should be Kh = 10 watthours and Kt=1.0 watthours.

The disk simulator displays for other meters are not all the same as this but the idea is the same. This particular meter also pulses a 25 ms IR ouput with each Kt pulse and you can pick it up with a sensor over the LED (which we use when making meter tests).

Great info!

One issue is that the blocks go blank when the unit is 'busy'. But since I am looking for a pattern of 10 to equal 1 rotation, I cam likely work around that.

Thanks!

FWIW, the Smart Meter specialist for our POCO, a very nice lady working on her master's in alternative energy, did not know this. She is supposed to call me back in a week with some answers to some questions I had for her. I will tell her what you told me and in the mean time, attempt a timed reading on my own meter.

 

[K8MHZ]

Mivey,

What does 'BUSY' actually mean when it is displayed?

Is that when the info is transmitted via wireless? Or does it mean something else?

 

[mivey]

Mivey,

What does 'BUSY' actually mean when it is displayed?

Is that when the info is transmitted via wireless? Or does it mean something else?

With the GE commercial/industrial meters it usually means it is communicating so that would be my guess. Some meters will flash all segments at the end of a rotation.

 

[gar]

121110-1423 EST

K8MHZ:

Just did a quick comparison of my Smart meter to my TED 1000. Present load was 1.180 kW per TED and appeared to remain constant during the test even though I was not watching TED. At the start and end TED still read 1.180 kW.

The smart meter went thru 3 cycles in 1 minute. Each cycle is 6 Wh. 18 wH per minute. 60*18 = 1.080 kWh, a 9% difference.

.

 

[mivey]

121110-1423 EST

K8MHZ:

Just did a quick comparison of my Smart meter to my TED 1000. Present load was 1.180 kW per TED and appeared to remain constant during the test even though I was not watching TED. At the start and end TED still read 1.180 kW.

The smart meter went thru 3 cycles in 1 minute. Each cycle is 6 Wh. 18 wH per minute. 60*18 = 1.080 kWh, a 9% difference.

.

I would expect the Smart meter to be at least 0.5% accurate.

 

[mivey]

Some meters will flash all segments at the end of a rotation.

Just looked at an Itron segment check flash and it can be set to no flash, 1 out of 8 seconds, or 1 out of 9 seconds with a preceeding 1 second blank.

 

[gar]

121110-1448 EST

mivey:

Without searching back thru various information I believe I remember the +/-0.5% for utility meters at unity PF and normal temperature. The major point is that TED is not a useful device for testing absolute energy consumption.

Even worse is trying to use TED to prove that power factor correction at the main panel will save a customer any money. Here under steady-state conditions one connects and disconnects the PFC capacitor. When I ran the test with a spinning disk meter there was no change, but TED showed a decrease in energy used with the capacitor connected.

.

 

[G._S._Ohm]

Don't forget, with an avg. branch circuit resistance of 0.5 ohm, at 10A this is 50w you're also paying for.

 

[fmtjfw]

My POCO is switching out to GE i210+C Smart Meters. I got mine the day before yesterday.

Yesterday I did a quick read and just got the feeling it was high. I keep pretty good tabs on my electric use. With the old mechanical meters I could time the rotation of the disc with a stop watch, do some math and get an 'instantaneous' value. I did that a few times, comparing the meter's value to a Kill-a-Watt connected to the only load on at the time and they were always identical.

Now, with no disc to time and no other value except accumulated KWh on the display, it's going to take more than a stopwatch and a calculator to determine accuracy.

Here is my dilema. As a learned electrician that has kept up on smart meters, I would consider such an observation (a gut reaction that the meter is reading high) from a customer to be a result of hype and not necessarily true. But also as a customer that has kept track of electrical use for over a year, I am now beginning to wonder. If I could do a quick test, which I no longer am able to, my fears could be laid to rest. This also means I cannot do a quick test for a customer at this point in time.

So, an Internet search led me to this product:

http://www.optimumstores.com/ted-5000-home-electricty-monitor-kits.html

Even with the above, since the Smart Meters only have a resolution of 1 KWh, a test would have to take a couple hours minimum to make a valid comparison.

So, back to my situation.

The last two October billing cycles (2011 and 2012) I used 310 and 336, respectively. Last December's billing was 389. So, let's take the 389 and divide by 30. I get 12.96, or let's say 13, per day.

Now, it's been less than 48 hours and my meter already reads 30, or 15 per day, and I am still short more than an hour to hit 48 hours.

What I am seeing is what any customer can see with no special equipment. The increase of 2KWh per day will certainly raise the price of the bill. Also, being 2 off out of 13 is a substantial percentage and worthy of further investigation.

For now, I am going to keep a log, as the meter is very easy to read. I have the usage history for a year on the Internet.

I also have a Kill-a-Watt to measure individual loads, just in case the meter is correct and something is using more energy than before. Right now it's on the fridge. It's been 18 hrs, and racked up 1.63 KWh. So there is 2.17. I still have over 12 to locate, and the fridge is the biggest load......except for maybe the computer?

The gist of the matter is, if I can't answer these questions for my own situation, how am I supposed to do it for a customer?

If all of a sudden their use increases by 15 percent and the only thing that has been changed is the meter, don't you think they will be looking for some answers?

As an electrician, if I were concerned about the POCO KWH meter, I'd just buy an old mechanical meter off the internet and put in an additional socket. If I were in a bad mood it with be outside beside their meter.