In 1999, JEA initiated a distribution automation (DA)program, focusing on its 26-kV distribution system for voltage control, fault isolation and service restoration. JEA (Jacksonville, Florida, U.S.) installed some 200 remotely monitored and controlled feeder reclosers in 90 26-kV feeders, and 88 DA-ready LTC voltage controllers at the substations. The utility also installed a package of DA software applications on computers in the control center for performing three major functions: distribution operation modeling and analysis (DOMA), volt-VAR optimization (VVO), and fault location, isolation and service restoration (FLIR).

The DOMA component of the software was based on the results of sets of unbalanced power flow calculations. The calculations were performed at least every 10 minutes where voltage profiles on the system were provided by the software. Calculations address:

• Load and voltage violations
• Lowest and highest voltage
• Highest voltage unbalance
• Most heavily loaded segment in the circuits of each distribution bus
• Margins to normal and emergency bus voltage limits
• Available real and reactive load reduction within given voltage tolerances.

The VVO software component recommends target settings for the LTC voltage controllers used on all 26-kV transformers. This target setting is sent to the field equipment via EMS/SCADA services.

The FLIR function determines the faulted section based on fault indications, and recommends switching sequences for fault isolation and service restoration through remotely controlled feeder reclosers. The automated fault isolation is implemented by strategically deploying automated reclosers (AR) with supervisory control and data acquisition (SCADA) on the circuits with the most faults. Service restoration is improved by adding SCADA-controlled switches to allow system operators to energize distribution segments that can be isolated using the AR’s information and SCADA services. The installation of remotely controlled reclosers showed immediate benefits in the improvement of SAIDI, greater flexibility in load balancing and maintenance activities, and more real-time information from remote sites beyond the substation fence.

The changes in distribution circuit connectivity, common within the system, are necessary for maintenance, load balancing and service restoration. The topology of the 26-kV circuits provides a variety of possible configurations; therefore, it is necessary when the configuration is changed that the VVO adjusts the voltage settings of LTC controllers to the new configuration. Note that the volt-VAR control function operates in a coordinated manner with other distribution operation processes.

At the outset of the DA program, the objective was to construct a real-time distribution model of the 26-kV system for estimating customer voltages at a preset periodic rate. The model results would then feed a program that would optimize the distribution bus voltage at each bus. The model became a major component of the program when analyzing electrical connectivity and electrical properties.

Electrical Connectivity and Electrical Properties The distribution models use existing data in the GIS system, including:

• Node connectivity from the distribution breaker to the secondary of distribution transformers.
• Conductor size, type and length between each node.
• All capacitors, including their size.
• All switches, reclosers, fuses and sectionalizers, showing their normal status of either open or closed.
• All distributions transformers with their size, impedance, tap position, summer and winter demand and monthly consumption. (All of these parameters are supplied for each phase.)
• All secondary meters and their summer and winter energy consumption.
• All primary meters and their three-phase peak demand.

All substation connectivity and installed connectivity devices are not in JEA’s GIS system and must be added to the model. Fortunately, these devices are static and require little maintenance. However, the substation connectivity can be complex with different switching options that can exist between the main bus and the secondary or transfer buses. Every possible transfer of a feeder circuit load to another feeder must be accounted for to properly address real-time operational switching that occurs for the purpose of maintenance on substation equipment.

Load Distribution
Because voltages along the distribution circuit depend on the distribution of load, it would be important to have a real-time telemetry of customer loads or to have an estimate of these loads. Because real-time telemetry is not yet practical for most utilities, a state estimation model is required. The model needs to consider all primary and secondary customers, expected scaling factors and load shapes. The scaling factor is derived from billing data obtained through an interface between the customer information system and the GIS.

The relationship between the billing data and the transformer associated with the relevant group of meters is checked for consistency, and the data-conversion program reports suspicious cases. Statistics show that around 5% of transformer kVA or loads are questionable. The state estimation routine corrects the data in accordance with predefined consistency rules. Based on the usual mix of customers, which include residential, small and large commercial, and industrial with unique usage patterns, the model could be complex and computer-intensive. Accordingly, the routine proportionately distributes the composite real-time load from SCADA at the feeder breaker in the substation to each distribution transformer based on the expected demand of each customer for the current time of day and the current seasonal load shape assigned to that group of customers.

Some customers have an additional feeder circuit serving their facilities, which could be used for backup service in case their primary service is interrupted via customer-owned switchgear. It also could be used for load sharing. However, without SCADA, sensitive computer modeling would be required to detect an increase in load on one feeder with a reciprocal decrease on the other feeder. In most cases, because the switch from the primary feeder to the backup is temporary, it is safe to assume the backup feeder is not used.

The Energy Management System and SCADA
The model and application software support all three phases of the circuit individually; therefore, the EMS must provide telemetry for all phases since a distribution feeder circuit should never be assumed to be sufficiently load balanced for each phase. Also important is the real-time representation of the switching/connectivity of the three-phase backbone. JEA has implemented a complete one-line graphical representation of all three-phase backbone distribution circuits, including all connectivity devices that can operate in the field or over SCADA for energizing or de-energizing segments of the system. If the device is not on SCADA, the dispatcher changes its status on the EMS system. The field personnel operate the device to keep the connectivity of the customer loads properly modeled.

In some cases, a device will be shown to be open when it is actually closed in the field, leading to de-energized zones in the model, which results in the model allocating the balance of the load to the remaining customers. However, if the device is shown to be closed when it is actually open, the model will allocate less load to customers.

These scenarios lead to inaccurate estimation of voltages and a resultant loss of control. To mitigate these problems, the modeling application includes a consistency check in load modeling that assumes normal loading will be between a minimum and maximum transformer loading use factor. If dividing the actual load at the breaker by the sum of all energized transformer capacities in the model produces a transformer use factor outside of these limits, the voltage control defaults to a predefined voltage set point for the LTC controller installed for the affected bus serving the feeder.

EMS alarm points and value limits should be implemented on all bus voltage telemetry to avoid taps on the transformer to run all the way to high or low settings. Constraints on the LTC controller, the EMS set point value and on the application logic provide assurance that runaway conditions are avoided due to flaws in the model. The use of any load telemetry from SCADA installed on the distribution circuit will enhance either the state estimation of the connectivity or the load distribution to the energized transformers between that node and the next node that has load telemetry.

Closed-Loop Voltage Control
With a complete connectivity model and a load model, the voltage at each customer load center can be estimated so that an optimum set point can be recommended for the LTC controller. This recommendation moves every customer toward a nominal 120-V service. Because some customers will be higher or lower than their nominal service, the algorithm uses customer load and penalty factors to derive a compromised set point. As a result, customers with large loads will have the most influence on the voltage setting. However, many circuits are made up of some large loads and many small loads, where the small loads will usually neutralize or override the few larger loads.

Once the voltage control application calculates a set point, it is sent to the EMS via a software data exchange interface and is subsequently transmitted to the output of the SCADA front-end, where it is delivered to the LTC controller via a substation RTU. At this time, the LTC controller will reset a timer and implement the new value. Because it is possible for the communications between the EMS and LTC controller to be interrupted, the controller will time-out if a “heartbeat” signal is not received in a predefined period of time. If the controller times out, it will default to a predetermined set point value that is programmed into its unit. The heartbeat signal is sent out every minute by the software application running on the EMS. If a connectivity device changes state either via SCADA notification or via a manual change by the dispatcher, the new status value is fed back to the voltage control application for an immediate recalculation of a new set point.

Other Uses of Voltage Control
Using voltage control to reduce demand on the system within customer acceptance is one objective the system operator can use. The operations engineer can set the desired minimum allowable customer voltage. The application environment will calculate potential real and reactive load reductions at different voltage reduction levels, which allows the operator to know how much load reduction is available at any given moment for various voltage reduction levels. The first level of voltage reduction is the point where any one customer experiences the lowest voltage as previously set by the operations engineer and could be as low as 114 V. The next level is an additional 2% decrease, which approximately totals a 5% decrease overall. The closer every customer’s voltage is to 120 V, the more load reduction for the system can be realized. Therefore, it is important to balance the voltages across the phases on a feeder circuit and across the feeder circuits on a distribution bus. For utilities that are competing in an open market, an additional voltage control objective would be to adjust the voltages based on the marginal cost of energy. If the marginal cost is greater than a predefined value, then a reduction in voltage would reduce the amount of high-priced energy to meet the demand. Adjusting system voltage to reduce load is practical and economical, because voltage reductions can be implemented with little or no impact on customer service when using a voltage-control application and the electrical model of the distribution system.

Voltage reduction has different results depending on the time of year and the makeup of the customer load. For example, JEA realized a significant decrease in load using voltage reduction during its highest peak, which surpassed its previous peak of 2732 MW with a new peak of 3166 MW. By using a 5% reduction in voltage, the load was reduced by more than 65 MW. There were no customer complaints, with all customers receiving service during this unprecedented peak. The amount of available load reduction via VVO is different under different operating conditions, including different loading of feeders and different circuit connectivities, affecting the voltage drop along the feeders. The DOMA application determines the available load reduction in real time, and the VVO application uses this availability when needed. A similar phenomenon occurs when reducing demand on the distribution system results in a rise in transmission voltage.

The Future of the Technology
The emergence of automated meter-reading technology can be the conduit for better information on customers’ loads, which would improve the accuracy of load modeling for individual distribution transformers. To use voltage measurements, the meters should be multifunctional with high accuracy. The current technology of retrieving data from thousands of meters does not meet the requirements for real-time use. Statistics collected from a limited number of meters can be used for better validation of the model. If meters are built to supply a voltage reading in addition to the standard energy and demand readings, high-speed communications will enable the development of applications that, as a minimum, could generate alarms through the DMS to the operators. These alarms could result in adjustments of the distribution model or additional validation of DA applications. It will be possible to determine the best tap position for each customer transformer based on statistical analysis over a period of time.

Summary
JEA has successfully implemented a systemwide combination of centralized and local voltage control, coordinated with other distribution operation processes. The benefits of the advanced voltage control, based on a real-time distribution operation model, include better power quality for customers, a better controlled load management process within predefined voltage-quality limits on customer sites, voltage and VAR support for the transmission system, and the possibility of limited load adjustment in accord with real-time pricing.

The process of setting up system models provides a better understanding of the requirements of the GIS database and better customer information. Being able to estimate the voltage in secondary load centers results in better customer service.

As utilities consider improving customer service, it is necessary to focus on their primary product, which is the voltage at the meter. The quality of this product can be understood through sophisticated models, and associated applications can be improved by integrating emerging technologies such as multifunctional AMR.

Donald Gilbert received the BSEE degree in 1982 from the Georgia Institute of Technology and has been engaged in engineering and management for the past 25 years. He first joined Georgia Power and then JEA, where he has been employed for the past 20 years. Gilbert has been involved with generation and transmission planning, as well as rate design. At JEA, he has been a project manager for EMS as well as a project manager for its SCADA system at its water plants and wastewater pumping stations. Gilbert is a member of IEEE and is a registered professional engineer in the state of Florida. GilbDC@jea.com.