What ComEd’s Outage Testing Revealed About its AMI Network

By Jeffry Birkmeier and Carolyn Johnson, ComEd

In connection with a pilot deployment of approximately 120,000 smart meters communicating through an advanced metering infrastructure (AMI) mesh network, Illinois utility ComEd and Silver Spring Networks (SSN) began a series of simulated outage tests to learn how to integrate this technology with the existing outage management system (OMS). In particular, ComEd wanted to know how the Silver Spring AMI system responded to various size outages and how the system could be utilized to more accurately assess outages and assist service restoration.

In evaluating the network’s outage detection capability, the goal was to determine:

  • Whether the number of received last-gasp outage notification messages sent by smart meters was sufficient for ComEd’s OMS (an ABB product: CADOPS) to generate an accurate outage assessment.
  • What the impact of meter hop count would be on outage notification. Hop count is the number of meters or other network devices that must pass along a last gasp before it reaches a network access or take-out point (the device that collects information from the meters and provides this to the ComEd back-end systems).
  • How long it took the network to recover from an outage and for restoration messages to be received by ComEd.
  • How the performance of the network varied by the extent or size of an outage.

In addition, ComEd was interested in determining if the number of relay devices with battery backup in the mesh network had a significant impact on the number of last gasps received during an outage and the time taken to restore the network following an outage. (All relays and take-out points in the ComEd network have battery backup.)

Testing Methodology

A cross-functional ComEd team including AMI pilot operations, distribution system operations and IT worked with SSN to design and execute a series of simulated outages to study the network’s performance in different outage scenarios. Because the AMI pilot is a production system, it was imperative there be no adverse or unintended network operation or customer effect with these simulated outages.

The selected testing area represented a segment of the ComEd service territory (flat, urban area) and a typical section of the AMI network, which is optimized for meter readings. SSN provided the technical expertise to simulate outages by transmitting messages to the network interface card (NIC) in each meter supplied with power from a specific electrical device (transformer, feeder, etc.) and telling the meter when to start and end the simulated outage.

Each test caused a last-gasp message to be sent at the beginning of the simulated outage and a full restoration cycle to be performed at the end, including a NIC reboot and the transmission of a restoration message. Tests encompassed various outage sizes, including single meter, transformer, fuse and feeder-level outages. Based on the timing and number of the last gasps received, the team manually created OMS tickets and sent those into the test OMS system to determine how well (and how quickly) it assessed the outages.

No customer experienced an electric service interruption during these tests and all tests were timed to avoid meter reading intervals. Approximately 4,000 meters were tested over a three month period, and the testing caused no operational issues. All tests were performed to simulate sustained outages; momentary outages (recloser operations) were not tested.

Outage and Restoration Results

The data gathered from each outage test was analyzed for the number of last gasps and restoration messages received, as well as the time it took to receive them. The table on Page 46 summarizes the number and types of outages simulated, the number of meters associated with each, whether the OMS system was able to accurately assess a given outage based on the last gasps received, the percentage of last gasps and restoration messages received, and the time it took to receive them. In addition to the simulated outages, the utility collected and analyzed data from several actual outages, the largest of which is included in the table as a “real outage.”

The hop count necessary to read each meter was also evaluated. This was done to determine whether the number of hops data must travel to reach a take-out point significantly affects the probability of receiving a last gasp. The results indicated that, for meters one hop away from a take-out point, 21 percent of the last gasps were received. Meters two hops to three hops away had the highest rate of received last gasps (32.5 percent), and that rate declined by only a few percentage points for meters with eight hops or nine hops. The lower receipt rate for one hop last gasps is most likely the result of radio frequency saturation based on the high number of last gasps to the access point in a very short time at the beginning of the outage.

These results were a surprising finding, as the underlying assumption was that the chance of receiving last gasps would decrease as the hop count increases, but it does not.

Observations and Conclusions

ComEd concluded that the AMI network optimized for consistent meter reads performed well in outage scenarios. Key findings include:

  • In all cases, when the last gasps were sent to the OMS system, the system correctly assessed each outage’s scope, regardless of the outage size. A noteworthy finding is that last gasps from all of the single-service outages were received. Also, enough last gasps from transformer, fuse and feeder outages were received to predict those outages accurately. For example, no fewer than five last gasps were received during any transformer outage. This provides some confidence that the system can accurately assess these types of outages even when a customer isn’t aware of them. These results, however, do underscore that a smart grid and smart meters do not mean that ComEd will receive an outage notification for every meter each time there is a power outage.
  • ComEd learned how last gasps are propagated, discovered that hop count doesn’t have a significant effect on outage notification and found that having a good distribution of last gasps is more important for outage assessment than sheer message volume. The rate of last gasps received proved sufficient for prediction in OMS, which indicates that building the AMI network for consistent meter reading and billing results in a density of battery backed-up devices adequate to predict outages.
  • More than 90 percent of the restoration messages were received within five minutes, which is much quicker than originally expected. Given that more than 30 percent of messages arrived in less than 30 seconds, it will be possible to rapidly identify when the grid has experienced a momentary outage as opposed to a true sustained outage. The complete restoration timing for the simulated feeder outages are show in the figure above.
  • These studies indicated that software (or another system) must be in place between the AMI network and OMS to filter the volume of last gasps in order to protect the OMS from being overwhelmed during normal operations (and particularly during storms). This software would need to filter out momentary outages, planned outages and other planned work such as meter exchanges, as well as possibly filtering the volume of last gasps by electrical connectivity (such as limiting the number of last gasps sent to OMS from the same feeder or transformer during a power outage). The AMI pilot and outage testing have helped ComEd compile a list of requirements to evaluate this software.

Next Steps

Having completed the first set of outage tests, additional outage testing is planned, along with exploring ways to leverage what the team learned to improve business processes, customer experience and overall grid reliability.

The next step in the outage testing efforts is to make each meter grid-aware by providing it neighborhood communication information and knowledge of the transformer, fuse and feeder or substation that supplies it. When it is grid aware, a meter will target its last gasp to a neighbor on a different transformer, fuse or feeder than its own. Different levels of grid awareness will be evaluated in meters to determine what effect this information has on the propagation of last gasps and restoration notices for different size outages.

Testing to determine the impact on outage response from a very dense AMI network, where each meter is essentially one hop away from a battery backed up device, is also under consideration. Additional relays with battery backup would be deployed, and tests would be executed to evaluate the impact on the receipt of last gasps and the timing of the restoration messages.

The gathering of a broad, cross-functional group, the articulation of clear goals and full management support were key to the testing program’s success. In addition, AMI partner SSN provided engineering and project management assistance every step of the way and remained very open to investigating their system’s behavior.

To develop an AMI/OMS integration strategy, a utility must realize that the smart in smart meter hinges more on what the utility does with the information than the technology itself.

Jeffry Birkmeier is ComEd’s business process manager, where he has worked in various positions for more than 17 years. Carolyn Johnson is currently an AMI operations analyst, and she has worked for ComEd for 10 years.

ComEd is a unit of Chicago-based Exelon Corp., one of the nation’s largest electric utilities. ComEd provides service for approximately 3.8 million customers, representing 70 percent of Illinois’ population, and is responsible for maintaining more than 70,000 miles of power lines that make up the electric transmission and distribution systems in northern Illinois.

More Power Engineering Issue Articles
Power Engineerng Issue Archives
View Power Generation Articles on PennEnergy.com
Previous articlePractical Advice n Developing a Smart Grid Work Force
Next articleAssessing the Smart Grid’s Data Security
The Clarion Energy Content Team is made up of editors from various publications, including POWERGRID International, Power Engineering, Renewable Energy World, Hydro Review, Smart Energy International, and Power Engineering International. Contact the content lead for this publication at Jennifer.Runyon@ClarionEvents.com.

No posts to display