An Affinity for Proper Connections

Ensuring Efficiency with Accurate Transformer and Phase Connectivity

By Robert Sonderegger, Itron

Power outages, whether caused by extreme weather or by local accidents, require a timely remedy by utility crews. Repairing distribution infrastructure following an outage could, however, result in the reconnection of some meters to a transformer or phase other than the one originally connected. In addition, there are situations, especially during power emergencies, where meter connections are rerouted without the utility’s records ever being updated. Over time, utility records and physical reality might diverge sufficiently, making it difficult to manage the distribution network.

Transformers may fail without warning because of overloading due to incorrect connectivity data on record, thus leaving all connected customers without power. Meanwhile, other transformers might be left inadvertently oversized with fewer meters than the original design specified. This causes unnecessary waste of equipment capacity and power. Theft detection strategies, based on comparing voltages of all meters connected to the same transformer, fail because knowledge of connectivity is faulty. Similarly, detecting high impedance connections-a threat to customer safety-is impaired by the lack of reliable data on connectivity.

Finally, outage detection and reporting systems rely on accurate knowledge of transformer and phase connectivity. In today’s communication networks, only a subset of the power-off-notifications (PONs) sent by all smart meters affected by an outage are “heard” by the head end. Precise connectivity information for each and every transformer is required to accurately and quickly determine the true extent and identification of all customers affected by an outage.

Mapping Transformer Connectivity

Traditionally, verifying transformer and phase connectivity has required either visual tracing of overhead lines or sending and receiving electrical signals over the wire. All traditional methods require considerable human resources.

Using robust machine learning techniques, patented algorithms have been developed to accurately determine meter phase connectivity and physical meter-to-transformer connectivity using voltage measurements from customers’ smart meters to determine an asset’s location.

These algorithms use the normally occurring voltage fluctuations at each meter as a signal to identify “friend vs. stranger,” in a manner of speaking. The voltage monitored by every meter changes as a consequence of one of three types of events:

  • Turning on or off a load at the premise that the meter monitors. Turning on an electric oven, for example, causes a slight decrease in premise voltage.
  • Turning on or off a load at a neighboring premise. If an electric oven is turned on in one premise, the slight decrease in voltage is felt by all meters connected to the same transformer.
  • Switching a device on the primary side of the transformer, such as a capacitor bank switching in or out, or a voltage regulator adjusting primary voltage.

The meter-to-transformer connectivity discovery method relies on correlating five-minute time series of voltage changes between any two meters within a user-settable limit, typically 1,000 to 3,000 feet. The correlation is conducted on a daily basis for a minimum of seven days, and the results evaluated to either confirm or reject the identity of each meter’s parent transformer. Where such identity is rejected, an alternate identity is indicated that is the more likely parent transformer.

Measuring Affinity

The advanced grid connectivity algorithms exploit the fact that a voltage change caused by switching a load at a premise is “felt” by most meters that are connected to the same transformer. By correlating each meter’s voltage changes to those of all other meters that are within a reasonable radius (approximately 1,000 feet), therefore, grid operators can distinguish meters on the same transformer that show the same, simultaneous voltage change, from meters on other transformers that do not show any simultaneous change.

Because interval data timing is imprecise, identifying actual connectivity requires collecting and correlating a sufficiently large number of voltage changes in order to use statistics-based approaches to minimize error. This method uses a quantitative measure, “affinity,” which characterizes the degree to which two meters experience the same local voltage changes. The degree to which affinity between any two meters varies helps in determining meter-to-transformer connectivity.

The orange curve in Figure 1 shows the distribution of affinities between meters connected to different transformers. The blue curve in this figure shows the distribution of affinities between meters on the same transformer. The number of combinations is shown on the left or right vertical axis respectively, and the affinity for each combination is shown on the x-axis. Note the nearly complete separation of affinity ranges between “same transformer” and “different transformer” meter pairs. For each individual meter, the analytic process calculates affinities with every other meter that utility records indicate is on the same transformer. The highest such affinity is called “home affinity.” For the same meter, affinities are then calculated with every meter not on the same transformer, within a configurable distance-such as 1,000 to 3,000 feet. The highest such affinity is called “away affinity.”

FIGURE 1: Distribution of Affinity of all meter connections under the same or under separate transformers
FIGURE 1: Distribution of Affinity of all meter connections under the same or under separate transformers

If home affinity is higher than away affinity, the utility-supplied connectivity data for the subject meter is validated as correct. If the away affinity is greater than the home affinity, the utility’s connectivity data is likely wrong for this meter. Furthermore, the transformer to which the meter with the highest affinity is connected indicates the most likely correct transformer for the subject meter.

Figure 2 shows a sample of 597 meters whose home affinity is shown on the Y-axis and the away affinity on the X-Axis. All meters in the top left half of the figure have correct connectivity as stated, while the 10 percent of meters in the bottom right half are clearly misconnected. For any one meter in this group, its away affinity on the x-axis points to the correct home for the subject meter.

FIGURE 2: Home vs. Away Affinity of 597 ServicePoints
FIGURE 2: Home vs. Away Affinity of 597 ServicePoints

Importance of Accurate Connectivity

Accurate meter-to-transformer connectivity as well as meter-to-phase is essential to enable a truly smart grid. It allows utilities to analyze each line segment from meter to feeder for revenue assurance and theft detection, and to gain valuable insight into their distribution systems. Utilities can detect power theft and high-impedance connections by comparing meter voltage to that of neighboring meters and find misplaced or misfiled meters.

Performing analytics on smart meter data without accurate and timely knowledge of meter-to-transformer or meter-to-phase connectivity is like finding your way with an out-of-date map. Advanced grid connectivity algorithms are like the GPS for smart grid analytics, allowing utilities to accurately and timely track meter locations in relation to the transformers that serve them.

Transformer Connectivity

Electric meters are connected to secondary distribution transformers to reduce distribution high voltage to safe levels for households. There are many transformers of different capacities along distribution lines, both underground and overhead. The choice of which transformer a meter is connected to is dictated by physical proximity as well as by optimal loading of transformer capacity. This connection between an individual meter and a distribution transformer is known as transformer connectivity.

Robert Sonderegger is a director and software engineering advisor at Itron. He is responsible for technical leadership, software innovation and advanced technology.

Previous articleXcel Energy plans 300 MW South Dakota wind farm
Next articleRecognizing a Tiger

No posts to display