Instrumentation & Controls, O&M

Putting the Proof to the Test: Case Study on Duke Energy’s Distribution Feeder Upgrade

Duke Energy has served the Raleigh, NC area for over 100 years, initially providing both electric and gas service as well as operation of electric streetcar transportation. Over the years, Duke Energy transitioned out of the gas and electric streetcar businesses, but continued to be the primary electric provider for the city.

This arrangement served the Raleigh business district well for over 30 years, resulting in a low frequency of events and power loss. Over time, as the number of customers grew, both load per customer and customer density increased as well. As greater portions of the modern economy have become reliant on dependable electric service, customer expectations for high reliability of that service have also grown.

Due to these factors, as well as a need to replace aging equipment in these areas, Duke Energy launched a proof of concept effort to better understand the roles automation and telecom technologies should play. The company’s objective is to develop greater expertise in these technologies through exploring opportunities to integrate high-speed automation switching solutions into its modernized distribution grid designs to improve electric service reliability for its customers.

Introduction to Proof of Concept

As part of a proof of concept for future distribution schemes, Duke Energy has completed the second phase of a project on a distribution system feeder for the Raleigh Central Business District underground system. The feeder consists of two radially operated 12kV underground circuits. Solid dielectric vacuum switches with integrated visible break were installed in nine network vaults during phase 1 of the project. To achieve high electric service availability for the central business district, a communications-assisted, high-speed protection system was developed. Its unique communication architecture utilizes IEC 61850 GOOSE messaging (the messaging protocol previously used by NASA’s Rover communications system for high interference communications) and serial based communications in parallel, enabling the relays to interrupt, isolate and restore power via the nine vault switches.

An important aspect of the acceptance test was testing the protection and control scheme. In this scheme, 18 relays and two communication technologies, work together as a system. Due to the interdependency of the network protection system and its components, it was critical to test every component as part of a system rather than limiting testing to a component level. Multiple acceptance criteria were defined by determining the initial state of the power system, the incident fault anticipated and the expected system state after protection system operation (i.e., after interruption, isolation and restoration of the system had occurred). The acceptance criteria were directly configured into the test environment by using a power system model that calculated the test set outputs. A single PC controlled a total of nine test sets (representing each of the nine network vaults), simultaneously injecting all signals according to the selected test case.

One requirement for placing the protection system into live operation after installation was the successful completion of field site acceptance testing (SAT). Site acceptance testing included testing the individual switching nodes during commissioning followed by a series of simultaneous network system response testing involving all of the switches.

The requirements for the proof of concept included the following:

1. System must be able to respond on its own to isolate an event.

2. System must have the ability to be flexible in its design to allow for the meeting of multiple

use cases for operation and circuit configuration.

3. System must have the ability to overcome the failure of primary systems, including

communications, switchgear, or automation relays.

4. System must be able to isolate a fault and restore the maximum number of customers within

a predetermined timeframe

5. Operation of the system must allow for either remote or local operation by an operator from

outside of the enclosed space environment to promote the safety of the employees.

6. System must allow for reconfiguration to its normal state with a single remote command.

7. System must allow for remote designation of new normal state.

8. System must be self-contained, not reliant on a single automation controller or other single

point of failure component.

9. Hardware design must allow for watertight conditions and the ability to isolate the control

from the switch components[EK1] .


Testing Methodology

To ensure an operable test setup, the test team utilized novel software which offered two features: the ability to run a power system simulation and the ability to control multiple test sets from a single software instance. Because the power system simulation introduced the ability to calculate currents and voltages, it simplified test case set up by calculating all currents and voltages for each relay in the power system automatically. Also, because the software enabled the team to control multiple test sets from one instance, the test case could run via a single button click. As part of the simulations, the software calculated the transient signals, distributed them to each test set and set the start time. After execution, all binary traces measured at the relay were transferred back to the software to be instantly assessed. One essential requirement of the simulation was a test of the circuit breaker response, which ran independently on the test set.  

The first execution injected a transient signal containing the fault incident. As expected, two of the breakers tripped selectively with a short delay. However, because the transient signals were already sent to all nine test sets, the test setup could not respond in real time. If the relays trip at the same time again when injecting the same fault quantities, the software automatically starts another iteration that will include the subsequent breaker events.

The same iterative process occurred again during the restoration.  The test sets measured a close command for the normally open breaker. The software recalculated the transients now containing the fault event, the isolation events and the restoration event. With the final execution, results similar to a real-time simulator were achieved.

The simplicity of test case definitions required for this iterative closed-loop simulation increased the likelihood of finding any errors in the protection system logic. In the case of a logic error, the misoperation is directly visible in a single line diagram, eliminating the need to investigate trip and close commands of ten different relays in a binary trace diagram.

The full loop system under test consisted of nine individual underground vaults located around downtown Raleigh. Each vault contained two relays. The primary relay measured two three-phase currents. The two three-phase voltages were measured via voltage sensors outputting low level signals. These conditions required each test set to have at least six phase currents and six low level voltage outputs.

Typically, when performing comprehensive field testing, a second test set would have been required to support the injection of test signals into the tap relay. For typical field testing of all nine vaults of the Raleigh underground system, 18 test sets would have been required. However, since the Duke Energy test team were primarily evaluating system behavior under full communication load (which mainly involved the primary relays), the testers determined that, as with the FAT (Factory Acceptance Testing), setting up an additional test set for each tap relay would be gratuitous. Instead, the test team opted for testing each tap way in the loop scheme, separate from the other tap ways, which still allowed for fully testing the system. This reduced the total setup actions required, thus reducing the risk of introducing error into the test sets.

Each underground vault test setup included one test set connected to a GPS antenna synchronized to IEEE 1588 precision time protocol (PTP). An Ethernet connection was used to communicate between these test sets via Duke Energy’s existing fiber network[EK2] .


Each test set needed to be connected to a relay inside the vault. In order to achieve this, a custom test cable was utilized to connect all required signals from the test set (switch status, CT secondaries, etc.) to the relay cabinet, effectively simulating the switchgear. The test cable used the same submersible connector and pin configuration as the cable that connected the interface cabinet to the relay cabinet.

All test cases were run on a single PC from a centralized location, above an underground vault in downtown Raleigh. The team performed two different groups of tests:

·  The first was a series of half-loop tests performed using five different test sets at four of the underground vaults. Each vault location had an assigned GPS clock for synchronization, with one location containing two test sets and sharing a single GPS clock. The two test sets were located at the vault where the tap way test would take place. This allowed for tests to be performed on eight of the primary ways and two of the tap ways, avoiding the need to remove the equipment from service.

·  The second group of tests (full loop tests) included all nine vaults in the system, utilized nine test sets and focused on testing all 9 primary ways together as a system.

Lessons Learned from Site Testing

Lessons learned to make setup smoother

The complexity of the system required the design team to coordinate up to nine test sets

simultaneously. This setup demanded significant resources (both equipment and personnel) to be on site for the testing in the individual vaults. Each test set required multiple connections for injecting the analog signals, binary signals and inputs from the relays, as well as connections for the GPS clocks. A simple connection error may have easily resulted in the test providing incorrect results, requiring

physical entry back into a vault to correct a connection. The team paid extra attention to ensure that all test connections were verified prior to the start of testing to avoid such adjustments during the site test[EK3] .

Lessons Learned from changes in FAT topology to SAT topology

Due to differences between the initial system design and the final system build, changes were required to the IEC 61850 GOOSE virtual bit mapping and serial bit mapping. Updates to relay logic were also made to reflect these changes. Due to discrepancies discovered during site acceptance testing, on-the-fly settings adjustments were required which introduced additional levels of uncertainty to the initial test results.

When designing the topology to be tested during the FAT, the team would have been wise to avoid assumptions about switchgear orientation relative to the source breakers. Had the team intentionally reoriented one or two devices in the FAT topology, a better understanding of the effects device orientation has on the logic would have been revealed. Necessary logic changes could have been

identified and implemented at the factory rather than during the SAT.

Locational Challenges

Due to the heavy foot traffic around each of the underground vault locations, the design team located the associated testing equipment within each of the nine vaults, to avoid having the equipment located on the ground level and assigning dedicated personnel to monitor each exposed access point. However, with the GPS clocks located in an underground vault rather than located in direct line of sight to open sky, the clocks experienced some losses of communication during the test set up. To remedy this, the team placed GPS clocks as close to the ground level as possible to allow for uninterrupted communication[EK4] .

About the authors: Peter Hoffman is Manager, Grid Monitoring, Control and Intelligence within Grid Architecture of Grid Solutions Engineering and Technology, an organization within Duke Energy’s Grid Solutions department. Grid Monitoring, Control, and Intelligence is focused on the technical aspects of effectively unlocking the value of future grid investments for customers and operations that are part of Duke Energy’s Grid Investment Plan, focusing on the development, evaluation, and enablement of both near term and strategic paradigm shifting solutions and capabilities within the operational technology space of the Distribution circuit, segment, and device levels. Peter has worked in several roles for Duke Energy, including Technology Evaluation Manager, Strategic Grid Business Investment Planning, Distribution Standards, Corporate Protection Engineer for Distribution, and various field engineering positions. He also has worked in Transmission Protection with the Tennessee Valley Authority. Peter is a registered professional engineer in SC and NC and is a graduate of NC State University with a bachelor’s degree in Electrical Engineering and an MBA with a focus on Innovation and Services Management. Peter and his family reside near Charlotte.

Erich Keller is Manager of Automation Engineering in Distribution Automation at G&W Electric Co. in Bolingbrook, IL. He is responsible for managing power system automation specification, design, factory acceptance testing and site commissioning. Before joining G&W in 2011, Erich was employed at ZIV USA in Des Plaines, IL. Erich Keller received a B.S. degree in Electrical Engineering from Valparaiso University and M.S. degree in Electrical Engineering from the Illinois Institute of Technology

Christopher Pritchard was born in Dortmund, Germany. He started his career in power as an electrical energy technician. Christopher received a diploma in Electrical Engineering at the University of Applied Science in Dortmund in 2006. He joined OMICRON electronics in 2006 where he worked in application software development in the field of testing solutions for protection and measurement systems and is now the responsible Product Manager for system-based testing solutions.

John Hart is a Senior Engineer in Grid Solutions at Duke Energy, focusing on protection and automation strategy development. He is a registered Professional Engineer and has worked in the energy industry for over 6 years. He received his Bachelor of Science in Electrical Engineering from University of North Carolina at Charlotte and is pursuing a Master of Science in Electrical Engineering from North Carolina State University.