Solid-state interval data recorders are the repositories of stories – the stories of the events behind them. We found errors at two metering points in the last year using interval data to discover mistakes and pinpoint the time they occurred. The first of these events was a mystery, a true “What, when, and how did it happen?” The second was straightforward –finding a loose component in a meter during a routine annual test. While the errors were embarrassing and costly, correcting them has been invaluable. So, this is also an account of managing mistakes to prevent their recurrence.
A generation and transmission utility that sells power wholesale to municipal and rural electric cooperative distribution utilities, the Lower Colorado River Authority (LCRA) also supports its customers with a key account program. At the wholesale level, LCRA delivers power to substations where it transfers ownership on the low side of the transformer. Unless the customer owns the transformer, LCRA’s Transmission Services Corporation (TSC) charges for peak demand during an 11-month period. On behalf of the Electric Reliability Council of Texas (ERCOT) grid, it also assesses a fee on all transformers’ average demand at the time of the ERCOT system’s peak in the four summer months, June through September. LCRA’s Generation Services group manages a key account program for participating customers, contracting with Texas Meter & Device of Waco, Texas, to manage participating customers’ poly-phase meters. TMD installs, tests, and verifies meter accuracy annually. Hunt Power, Arlington, provides meter data services.
That background provides the setting for these two events.
Let’s begin with the mysterious load drop in a municipal utility. The first person to notice something awry was a transmission billing analyst who was developing annual estimates of peak loads for wholesale electric customers’ transformation and transmission billing determinants. He graphed each wholesale customer’s total demand by month and saw that the peak demand for one municipality had dropped approximately 20 percent in a year. He started asking questions – “Did anyone know of a business or major building that had left?” Customer service representatives verified that the community had experienced no major changes.
An analyst in the generation group studying the data developed a simple but effective test using interval data. He compared 15-minute intervals for two years and graphed their relationship in terms of a ratio around a horizontal line of their mean. Looking at the intervals’ ratio, he saw that the load began to decline in the fall of 2004. The drop was neither immediate nor steady. More analysis identified that from mid-November 2004 through September 2005, the city’s metered demand had decreased from about 2.4 megawatts (MW) to 1.8 MW. At first field staff thought the city’s upgrade of its distribution system from its 4,160-volt service to a 12,500 may have caused the change. The municipal’s electric superintendent dispelled that theory, noting that the line had not been built. When an LCRA meter technician re-tested the meter in October 2005, he found that the shield wires surrounding the conductors from the regulator to city breaker were grounded at both ends, causing current to circulate through the meter’s current transformers.
The problem stemmed from temporary service LCRA built for the city’s distribution improvements. While crews installed new transformers and controls for the 12.5 kilovolt (kV) service, another group set temporary service adjacent to the permanent substation. Temporary equipment included a step-down transformer, breakers, conductor cable insulated and shielded for use on the ground, metering current transformers (CTs), and the meter with its instruments.
Proper installation of shielded cable calls for grounding only one end. In this case, the crew connected the grounds at both the regulator and breaker, causing current to circulate in both directions. The CTs see current flowing in one direction, record it, and then when current flows back, they subtract the reverse current, leaving a remainder that is less than delivered power. When the temporary equipment was set, the meter tech had followed standard practice and probed the conductors above and below their CTs but between the grounds. The readings matched, so the installation would appear to be proper.
With a shielded cable, the correct method of measurement is to probe the conductor where it is not shielded, at both the bus and relay ends. That measurement compared with a reading at the meter shows the amperage as delivered and measured. A difference between amps at the bus and at the meter would indicate a problem in the connections or grounding; for example, the meter check on the improperly grounded conductor showed 171 amps at the connections but only 147 at the CT. The grounds between the regulator and CT accounted for the difference.
Merely finding and correcting the mistake addresses only the immediate problem. To turn errors into opportunities for education, LCRA TSC’s field managers have established a formal process for documenting mistakes, reviewing their causes, recommending improvements, and not laying blame except in instances involving safety violations or negligence. As the manger for maintenance services, Peter Larkam, said, “You don’t want to absolve people from the responsibility of doing something right.” LCRA wants to encourage its workers to report mistakes and cite corrections. To assist with this effort, LCRA TSC has engaged a consulting firm that specializes in industries where human error may lead to catastrophic results: nuclear power, hospitals, and airlines, to advise on this process. Completed studies form a library for supervisors and managers to review in order to avoid repeating mistakes. Additionally, reports are analyzed for trends that may not be apparent when examining only isolated events.
Larkam chartered a temporary team to report on the error at the municipal substation. Its key recommendations were to change the meter tests for temporary service to measure current at the regulator and breaker to ensure readings match and to develop written procedures for installing temporary substations. Those steps are at the construction and installation end. It also charged the recipients of meter data, analysts in generation services, to look at interval data for all sites after meter tests or new meters are installed, using the ratio comparison developed while looking for the lost kilowatts.
At the consumer level, the cause of the meter error occurred inside the meter. The sensor for the A Phase current transformer pulled away from its internal terminal during the meter’s annual test in January 2005. No one knows how the Molex connector came unplugged. In the 2005 test, the technician checked the potential and current transformers for condition and accuracy with a phase-angle test device called a “Bird Dog.” Next he removed the meter from its “A-Base” mounting, meaning it is hard-wired in place. Meters in sockets have to be pulled from their mountings; a sudden and hard tug could have loosened the connector. In this case, the tech unscrews leads, lifts the meter, and wires into bench-testing equipment in the van. All test results for 2005 came back good. The meter tech re-connected the meter into its bracket, restored power, and waited for Hunt Power to read the meter to get the first 15-minute interval. With the communication link working and the data file verified, the TMD tech completed the on-site test.
Yet sometime between bench test and mounting the sensor came loose.
During the next year’s bench test, the meter was running 30 percent slow. The technician opened the cover and found the sensor hanging inside. He replaced it, and the ensuing test indicated the meter was working properly. The loose sensor caused the meter not to record current from the A-phase conductor, approximately a third of the customer’s load for a year.
Even though interval data did not play in the discovery of this error’s root problem, the test technique of ratio comparison verified date and time, as did the data on the meter channels recording amps and volts by phase. When our MV-90 operator learned about the loose sensor, he asked the analyst who developed the ratio comparison to look at the customer’s data for 2004 and 2005. A few minutes later, the analyst returned and asked, “What happened on Jan. 21 last year at about 1:30 in the afternoon?” The graph’s curve comparing the two years’ intervals pegged the time the sensor failed. A second analyst looked at channels recording amps and volts and saw the A-phase registering zeroes from the time the meter went back into service. Any doubt about cause or time disappeared.
Next, we followed the example of Transmission Services. We reported on cause and recommendations, and again, improvements start at the site. First and most important, TMD has added a step for new installations and tests. After all other tests, the tech changes the face display by putting a magnet above a reed switch on the meter’s circumference. The switch opens and changes the visual display from engineering units to volts by phase, verifying that all phases are operating properly. After meter tests, an analyst compares the meter’s data to periods before the test – day, week, and week a year earlier. In addition, the checker looks at volts and amps if the meter is programmed to record them. Last, and farthest from the site, the programmers of our billing system have created a high-low report that we run at billing time.
About the Author
Tom Knutsen is an energy services consultant for the Lower Colorado River Authority (LCRA) in Austin, Texas. He and three other co-workers share responsibility for the meter and communications infrastructure, data collection, and billing data delivery for key end-use consumers served by LCRA’s wholesale customers. His writings on interval data management have been published in Transmission and Distribution World and Electric Energy T&D. Before LCRA, he worked five years as a reporter for daily newspapers in Texas. He holds an A.B. in history from Yale University and a master’s in communication from The University of Texas at Austin.
E-mail: tknutsen@lcra.org
A generation and transmission utility that sells power wholesale to municipal and rural electric cooperative distribution utilities, the Lower Colorado River Authority (LCRA) also supports its customers with a key account program. At the wholesale level, LCRA delivers power to substations where it transfers ownership on the low side of the transformer. Unless the customer owns the transformer, LCRA’s Transmission Services Corporation (TSC) charges for peak demand during an 11-month period. On behalf of the Electric Reliability Council of Texas (ERCOT) grid, it also assesses a fee on all transformers’ average demand at the time of the ERCOT system’s peak in the four summer months, June through September. LCRA’s Generation Services group manages a key account program for participating customers, contracting with Texas Meter & Device of Waco, Texas, to manage participating customers’ poly-phase meters. TMD installs, tests, and verifies meter accuracy annually. Hunt Power, Arlington, provides meter data services.
That background provides the setting for these two events.
Let’s begin with the mysterious load drop in a municipal utility. The first person to notice something awry was a transmission billing analyst who was developing annual estimates of peak loads for wholesale electric customers’ transformation and transmission billing determinants. He graphed each wholesale customer’s total demand by month and saw that the peak demand for one municipality had dropped approximately 20 percent in a year. He started asking questions – “Did anyone know of a business or major building that had left?” Customer service representatives verified that the community had experienced no major changes.
An analyst in the generation group studying the data developed a simple but effective test using interval data. He compared 15-minute intervals for two years and graphed their relationship in terms of a ratio around a horizontal line of their mean. Looking at the intervals’ ratio, he saw that the load began to decline in the fall of 2004. The drop was neither immediate nor steady. More analysis identified that from mid-November 2004 through September 2005, the city’s metered demand had decreased from about 2.4 megawatts (MW) to 1.8 MW. At first field staff thought the city’s upgrade of its distribution system from its 4,160-volt service to a 12,500 may have caused the change. The municipal’s electric superintendent dispelled that theory, noting that the line had not been built. When an LCRA meter technician re-tested the meter in October 2005, he found that the shield wires surrounding the conductors from the regulator to city breaker were grounded at both ends, causing current to circulate through the meter’s current transformers.
The problem stemmed from temporary service LCRA built for the city’s distribution improvements. While crews installed new transformers and controls for the 12.5 kilovolt (kV) service, another group set temporary service adjacent to the permanent substation. Temporary equipment included a step-down transformer, breakers, conductor cable insulated and shielded for use on the ground, metering current transformers (CTs), and the meter with its instruments.
Proper installation of shielded cable calls for grounding only one end. In this case, the crew connected the grounds at both the regulator and breaker, causing current to circulate in both directions. The CTs see current flowing in one direction, record it, and then when current flows back, they subtract the reverse current, leaving a remainder that is less than delivered power. When the temporary equipment was set, the meter tech had followed standard practice and probed the conductors above and below their CTs but between the grounds. The readings matched, so the installation would appear to be proper.
With a shielded cable, the correct method of measurement is to probe the conductor where it is not shielded, at both the bus and relay ends. That measurement compared with a reading at the meter shows the amperage as delivered and measured. A difference between amps at the bus and at the meter would indicate a problem in the connections or grounding; for example, the meter check on the improperly grounded conductor showed 171 amps at the connections but only 147 at the CT. The grounds between the regulator and CT accounted for the difference.
Merely finding and correcting the mistake addresses only the immediate problem. To turn errors into opportunities for education, LCRA TSC’s field managers have established a formal process for documenting mistakes, reviewing their causes, recommending improvements, and not laying blame except in instances involving safety violations or negligence. As the manger for maintenance services, Peter Larkam, said, “You don’t want to absolve people from the responsibility of doing something right.” LCRA wants to encourage its workers to report mistakes and cite corrections. To assist with this effort, LCRA TSC has engaged a consulting firm that specializes in industries where human error may lead to catastrophic results: nuclear power, hospitals, and airlines, to advise on this process. Completed studies form a library for supervisors and managers to review in order to avoid repeating mistakes. Additionally, reports are analyzed for trends that may not be apparent when examining only isolated events.
Larkam chartered a temporary team to report on the error at the municipal substation. Its key recommendations were to change the meter tests for temporary service to measure current at the regulator and breaker to ensure readings match and to develop written procedures for installing temporary substations. Those steps are at the construction and installation end. It also charged the recipients of meter data, analysts in generation services, to look at interval data for all sites after meter tests or new meters are installed, using the ratio comparison developed while looking for the lost kilowatts.
At the consumer level, the cause of the meter error occurred inside the meter. The sensor for the A Phase current transformer pulled away from its internal terminal during the meter’s annual test in January 2005. No one knows how the Molex connector came unplugged. In the 2005 test, the technician checked the potential and current transformers for condition and accuracy with a phase-angle test device called a “Bird Dog.” Next he removed the meter from its “A-Base” mounting, meaning it is hard-wired in place. Meters in sockets have to be pulled from their mountings; a sudden and hard tug could have loosened the connector. In this case, the tech unscrews leads, lifts the meter, and wires into bench-testing equipment in the van. All test results for 2005 came back good. The meter tech re-connected the meter into its bracket, restored power, and waited for Hunt Power to read the meter to get the first 15-minute interval. With the communication link working and the data file verified, the TMD tech completed the on-site test.
Yet sometime between bench test and mounting the sensor came loose.
During the next year’s bench test, the meter was running 30 percent slow. The technician opened the cover and found the sensor hanging inside. He replaced it, and the ensuing test indicated the meter was working properly. The loose sensor caused the meter not to record current from the A-phase conductor, approximately a third of the customer’s load for a year.
Even though interval data did not play in the discovery of this error’s root problem, the test technique of ratio comparison verified date and time, as did the data on the meter channels recording amps and volts by phase. When our MV-90 operator learned about the loose sensor, he asked the analyst who developed the ratio comparison to look at the customer’s data for 2004 and 2005. A few minutes later, the analyst returned and asked, “What happened on Jan. 21 last year at about 1:30 in the afternoon?” The graph’s curve comparing the two years’ intervals pegged the time the sensor failed. A second analyst looked at channels recording amps and volts and saw the A-phase registering zeroes from the time the meter went back into service. Any doubt about cause or time disappeared.
Next, we followed the example of Transmission Services. We reported on cause and recommendations, and again, improvements start at the site. First and most important, TMD has added a step for new installations and tests. After all other tests, the tech changes the face display by putting a magnet above a reed switch on the meter’s circumference. The switch opens and changes the visual display from engineering units to volts by phase, verifying that all phases are operating properly. After meter tests, an analyst compares the meter’s data to periods before the test – day, week, and week a year earlier. In addition, the checker looks at volts and amps if the meter is programmed to record them. Last, and farthest from the site, the programmers of our billing system have created a high-low report that we run at billing time.
About the Author
Tom Knutsen is an energy services consultant for the Lower Colorado River Authority (LCRA) in Austin, Texas. He and three other co-workers share responsibility for the meter and communications infrastructure, data collection, and billing data delivery for key end-use consumers served by LCRA’s wholesale customers. His writings on interval data management have been published in Transmission and Distribution World and Electric Energy T&D. Before LCRA, he worked five years as a reporter for daily newspapers in Texas. He holds an A.B. in history from Yale University and a master’s in communication from The University of Texas at Austin.
E-mail: tknutsen@lcra.org
