Voltage analytics has been proven to quickly and efficiently stop power diversion by comparing the difference in the relationship between voltage and kilowatts per hours (kWh) in a properly functioning meter to that in a meter with a diversion or other problem. A few utilities have been implementing voltage analytics for the past several years, but they have been limited to using the approach manually. This has been very slow and time-consuming, and generally conducted on a transformer-by-transformer basis rather than in any sort of holistic way across the entire infrastructure.
Meanwhile, diverted loads that are often hidden behind walls or buried below ground go undetected and continue unabated. Worse, diverted wires with no secondary voltage protection are extremely hazardous — any type of short on the utility side of the meter has the potential to arc and continue to burn, since there are no breakers for auto shut-off. This dangerous lack of visibility can be solved by automating the voltage analytics process to deliver faster results and a more comprehensive view, while enabling voltage analytics to scale across any size of utility. Perhaps most importantly, this model ensures that problems are identified in near-real-time, before they impact revenues or, worse, create a safety risk that leads to injury or death.
How Voltage Analytics Works
Unlike software solutions that detect diversion based on usage patterns, voltage analytics compares the difference in the relationship between voltage and kWh in a properly functioning meter to that in a meter where there is a diversion or other problem. The technique was proven over the five-year period between 2012 and 2017 at one California utility, where its use revealed that 750 residential customers had diverted loads, or approximately .7 percent of the ratepayer base. The average diverted load represented a billing loss of $5,000 per month, and the utility captured over $2.6 million per year in revenue loss from these diverted loads. They have also realized a 3 percent reduction in T&D losses between 2012 and 2019. Another small California utility with less than 100,000 residential customers has identified 450 diverted loads using voltage analytics. A larger investor-owned utility has identified thousands of diverted loads using these techniques.
There are several types of issues that voltage anomalies can identify. These include loads connected on the line side of the secondary (diverted loads), meters that have been mapped to the incorrect transformer, and faulty utility-side electrical connections.
Voltage analytics detects these problems based on a relatively simple concept: when the voltage drops at one meter as compared to the average of all meters on the same transformer, it is very likely that the meter is being bypassed or there is some other problem, if there is not simultaneously an increase in wattage (see Fig. 1).
Fig. 1: A drop of as little as a half volt at one meter as compared to the average of others on the transformer will expose a diversion or other problem.
The same voltage anomaly also can expose meters that have been mapped to the incorrect transformer, and those with faulty utility-side electrical connections. While each is important, fraud is the most expensive problem and poses the greatest safety risks. Diverted wires with no secondary voltage protection are very dangerous — everything is accomplished from inside the home or garage, without pulling the electric meter, by cutting open the sheetrock and entering the back side of the meter panel to expose the utility’s line-side service wires. Voltage analytics can identify opportunities to detect these hidden diversions and prevent fire or electrocution at the meter.
The connections associated with diverted loads are most often made outside of any type of safety electrical box. Additionally, diversions mask the true kVA the associated transformer is serving. The large sizes of these loads can cause transformers to overheat without distribution staff being aware, leading to unplanned outages during peak loading periods.
Using voltage analytics to expose large diverted loads and faulty secondary connections also helps prevent damage to ratepayers’ connected electrical equipment. It is not uncommon for the larger diverted loads such as marijuana grows to cause extreme sags in voltage of up to six volts or more. These unmetered loads can run for 20 hours a day or more, and the associated voltage sags will damage the utility side electrical connections to homes fed from same service box. They cause flickering lights and, eventually, these extreme voltage sags can cause overheating and failure of inductive loads.
It is not uncommon for diverted wires to be left connected when a home is sold or rented to an unsuspecting new homeowner/tenant. The connected “hot” wires are covered up by patching sheetrock where diverted loads previously were connected. The diverted wires that no longer have connected loads pose a serious risk of injury or death to unsuspecting person who may uncover these diverted wires.
One of the biggest challenges in these situations is that there must be a diverted load in order for voltage analytics to expose the diversion. If the utility doesn’t catch the diversion before it is removed, the diverted wiring will not be detected using analytics. It is critical to run voltage analytics continuously, grid-wide, to avoid any gaps in visibility during a service interruption at any location where someone might uncover a hidden diversion and risk injury or death.
Automating Voltage Analytics
The first step to launching an automated voltage analytics program is to build filtering and correlation algorithms that can be run across the extremely large data sets of a typical advanced metering infrastructure (AMI) system. These algorithms are based on roughly a dozen voltage interrelationships among and between meters and transformers that are the biggest indicators of safety or theft issues. A combination of correlations, exceptions and iterative queries all change continuously as meter/transformer interrelationships and diversion techniques evolve.
Once the analytics foundation has been built, meter data is gathered and analyzed to identify which need additional, physical investigation. A database is first set up, and an automated loading process is established. Next, the AMI is rid of data irregularities that hide fraud. This includes reviewing data about how meters are mapped to transformers, and correcting the mapping, where needed, to unmask and accurately identify energy diversion. The typical mid-sized utility can have thousands of these mapping errors.
The next step is to find and correct faulty connections, which also can mask or misrepresent energy diversions. Once the data has been sanitized, voltage analytics can find theft more quickly, based on much smaller anomalies (see Fig. 2), while also exposing utility-side connection issues and improving the transformer voltage and load management process.
Fig. 2: Voltage drop identified within two days of diverted load being installed, after cleaning up system data and launching continuous monitoring and analytics.
The analytics process is applied, section by section, across the entire grid with the goal of “touching” each meter on, say, a monthly, quarterly or semiannual basis. Meanwhile, there is close collaboration and a feedback loop with technicians who visit suspected theft sites and report findings, to ensure that algorithms can be kept current. The process is scalable to any size utility, and any frequency of meter review.
There are other types of residential theft that utilities can identify using voltage analytics combined with other data, including:
- Air conditioning and pool pump loads that are running off diverted loads.
- Marijuana grow house diversions that are revealed not through voltage anomalies but through time-clock signatures and/or LED lighting signatures.
- Single-fed meters or rural accounts where less than 3 meters are served by a single transformer
- Diversions that are implemented through installation of a resistor across the secondary of the meter’s current transformer.
Various types of non-residential theft can also be identified with voltage analytics, including large motors, irrigation pumps and commercial three-phase meters that are running off diverted loads.
Optimizing Grid Integrity and Efficiency
World Bank has estimated that it is 3 times cheaper to save 1 kWh by improving efficiency than investing in new generation plants. Voltage analytics offers benefits here, as well. Beyond ROI from theft detection, adding voltage analytics to a utility’s technology toolkit also improves grid safety, integrity and efficiency. These benefits are paid for through the revenue recovery that is made possible because of the technique’s high rate of success in identifying diversions.
Voltage analytics helps optimize the integrity and efficiency of existing grid infrastructure in two key ways: improving transformer voltage and load management, and correcting mis-mapping of meters to transformers.
When a transformer has a large unmetered load, such as when it is serving a marijuana grow, its duty cycle can run upwards of 95 percent. Having additional unmetered loads on the same transformer can cause the duty cycle to run high enough to exceed its kVA rating. By helping to identity unmetered loads and alert utilities when transformers are at risk of exceeding their kVA limit, voltage analytics can help ensure more accurate transformer load management. They can more effectively evaluate transformer duty cycles and loading, and pre-empt transformer overuse, overheating and unplanned outages.
The second big benefit of voltage analytics for grid integrity is that it allows meter-to-transformer mapping errors to be Identified and corrected. This, too, is essential for transformer load management. In one case, a utility applied voltage analytics to its smart meter data and found thousands of meter-to-transformer mapping errors. Mapping errors can also mask diversions and other faulty connections, as shown in Fig. 3.
Fig. 3: The voltage and kWh for this transformer show a mapping issue based on the distinct differences between Group 1 and Group 2
The value and ROI of voltage continues to grow as voltage analytics findings are incorporated into updated queries that improve the process. It has been estimated that a mid-sized utility can generally project from 200 percent to 600 percent in annual returns by investing in automated voltage analytics to recover past, unpaid usage fees. Meanwhile, they are eliminating the costs of not implementing the approach, including injuries, liabilities and the future revenue losses associated with letting unmetered loads continue, unabated.
Source: energycentral
