More buzz today about smart meter "failures", this time out of Texas. (Link to New York Times report, but hat tip to The Dallas Morning News.)
Following the pattern of Pacific Gas and Electric Company's problems in Bakersfield, Oncor now has a state senator nipping at their heels asking for an independent audit to verify the accuracy of meters being installed in their service area.
I happen to have a good friend in the utility industry who is responsible for automated meter technology and systems (though only on the natural gas side, not the electric meters), so I've gotten a birds-eye view of the challenges that have occurred for large-scale deployments.
We've had great discussions about the problems of "newly high reads" after meters are changed out, and can point to several possible failure modes that would explain higher reads (and bills!) beyond the usual utility reasons of rate increases and weather conditions.
First, all of the failure modes for electric meters cause under-registration of usage - there is simply no case where a meter will register falsely high usage, absent a meter constant error. (Meter constants are introduced for commercial and industrial services, where potential and current transformers are used to send say one hundredth of the electricity through the meter. Sometimes the constant is misapplied.)
As mechanical meters age, the mechanisms wear and introduce drag on the register, gradually making them run slower. Newer solid state meters don'e have this problem, but the vast majority of installed meters are mechanical models that essentially operate like electric motors.
In rarer cases an element that measures use on one leg of a polyphase meter can fail, cutting registered usage by a third or half. In either case, a meter replacement will properly measure use that can be somewhat or significantly higher than historical readings.
The next case is one that utilities don't particularly like to talk about too much: a fair number of reads are estimated. This is usually done when there are access issues - the meter reader simply can't get to the meter to take the read. So, an estimate is made based on prior use history - sometimes with poor accuracy if there have been changes at the customer premise.
In some cases, for some customers, reads can be estimated for months in a row, with the expectation that a "true-up" read can be taken at some point.
If the estimated reads are way out of line with actual use, the true-up read (which always occurs with a meter upgrade or change out) can be a "shocker" to the customer.
The estimated read/true-up case can also be extended to customers who read their own meters, which is a very common arrangement in both dense urban and remote rural areas. Customers post a mock display of the reading in a visible location for the meter reader, or mail in a post card. Nominally, the utility requires access to the meter annually for a true-up/verification, but can be lax in fulfilling that obligation.
Why do utilities point to weather (it was hotter or colder in the month we installed smart meters) or rate increases instead of these scenarios? They simply don't want to acknowledge that in some cases meters are inaccurate (though always under-reading), and in others they aren't directly reading them. Those are messages that regulators and customers don't want to hear.