In the last decade, volt/volt-ampere-reactive (VAR) optimization became one of the most desirable applications implemented on the distribution system. Primary objectives of volt/VAR control include power loss minimization, known as VAR control, and energy demand reduction, known as voltage control or conservation voltage reduction (CVR), or some combination of the two.

This control methodology uses load tap changers (LTCs), voltage regulators (VRs) and capacitor banks to achieve the desired objectives. The application consists of two main parts — voltage flattening and voltage reduction. Field capacitors are normally used to flatten the voltage along the feeder, while LTCs and VRs are normally used to reduce the voltage.

Volt/VAR Control at BGE

Baltimore Gas and Electric (BGE) developed a smart grid distribution system pilot project on six test feeders in 2011 with three main objectives:

  • Implement volt/VAR control to reduce the energy consumption

  • Install capacitor bank controllers with two-way communications to improve the reliability of field capacitor

  • Reduce the customer average interruption duration index by installing fault circuit indicators.

Unlike the majority of utilities that use LTCs or VRs and capacitor banks for voltage and VAR control, BGE uses only feeder capacitor banks in most cases. Current volt/VAR control is achieved by maintaining the bus voltage at predetermined levels as a function of the substation transformer loading. Capacitor banks are switched first on the feeder most in need of reactive power correction. This approach maintains proper voltage and keeps the power factor within the desired limits. In addition, the control enables faster voltage response to load changes during peak loading and lower voltage levels during the times with low loading.

BGE's smart grid distribution system pilot project uses a volt/VAR control optimization system developed by Cooper Power Systems, thus leveraging the existing Yukon capacitor voltage control system. BGE identified five pilot feeders for the implementation of the new volt/VAR control.

In order for volt/VAR optimization to work, it is necessary to have either LTCs or VRs in the substation. As BGE does not use either of these devices, new Cooper VR-32 voltage regulators have been installed on five pilot feeders along with new CL-6B regulator controllers. One of the three new controllers on each feeder is equipped with a Raven X modem, which communicates to the Yukon server through the Verizon 3G network. Communication with the remaining two regulator controllers on the feeder is accomplished by using the fiber-optic loop between the three controllers. In addition, new two-way Cooper CBC 7024 capacitor bank controllers along with Raven X modems have been installed on each of the capacitor banks.

To capture end-of-line (EOL) voltages, approximately 40 Landis+Gyr FOCUS AX smart meters have been installed for residential customers and 12 SATEC EM920 eXpertmeters communicating DNP3 protocol have been installed for industrial and commercial customers. Landis+Gyr smart meters are equipped with two-way cell code division multiple access (CDMA) network interface cards, while SATEC EM920 meters have a general packet radio service (GPRS) modem, in order to provide a real-time voltage feedback to the volt/VAR optimization application.

The Yukon volt/VAR application monitors real-time voltage, watts and VARs from LTCs, regulators, capacitors, medium-voltage sensors and additional monitoring points such as customer meters. Using this real-time set of analog measurements, the application triggers a control period during which the real-time power factor and voltage measurement set is assigned an operational cost. The operational cost is determined from the analog measurement set compared against substation power factor and voltage targets. The objective of the application is to minimize the operational cost by managing real-time power factor and voltages as close as possible to the substation power factor and voltage targets. The scheme combines both VR and power-factor correction in real time, ensuring optimum performance.

Internally Developed Volt/VAR Control

The BGE distribution system has few LTCs and VRs, and, in most cases, capacitors are used for volt and VAR control. Given that, volt/VAR optimization that involves voltage flattening and voltage reduction cannot be implemented. For the purposes of the pilot project, in addition to five test feeders with Cooper volt/VAR optimization, BGE decided to use one additional residential feeder and apply internally developed volt/VAR control algorithms that only use capacitor banks, and compare the energy savings to the current volt/VAR optimization.

The residential test feeder has seven capacitor banks equipped with the new two-way Cooper CBC 7024 capacitor bank controllers. In addition, 10 SATEC EM133-SP2S meters, communicating DNP3 protocol, have been installed at selected locations to measure EOL voltage. The meters use a Raven X 3G CDMA modem and have an adapter that attaches to the residential 2S meter base.

As the substation that contains the test feeder is equipped with a Motorola MOSCAD remote terminal unit, a Raven X modem was added to the SEL-351 relay from Schweitzer Engineering Laboratories (SEL) within the substation to retrieve the data every 15 seconds for the algorithm to run. The algorithm itself was programmed into the SEL-3354 device by using SUBNET Solutions' application. This device also was used as an open process control server for storing the data needed for measurement and verification in OSIsoft's PI Historian, as well as an open process control client for reading the distributed automation switch status points from the PI Historian to know the connectivity model and disable the volt/VAR optimization if the feeder is reconfigured.

Capacitor-Only Volt/VAR Control Optimization

The capacitor-only volt/VAR control optimization minimizes the average voltage on the feeder while still keeping the customer voltages and power factor within the desired limits. The user inputs to the algorithm are minimum and maximum voltages on the capacitor banks, desired power factor, bandwidth and three user-defined EOL voltage levels. EOL voltages are measured on the customer premises, with the lowest voltages generally at the end of the feeder.

In addition, the algorithm sets the limits for the number of daily operations of a particular capacitor bank, checks the quality of the reported voltages, prevents the failed capacitor banks from switching and prevents the capacitor banks from consecutively operating within a certain time frame. The application contains a voltage drop and rise table, which describes the voltage changes at the substation bus and all capacitor banks as a function of switching each of the capacitor banks on the feeder. The table has been developed based on the existing feeder model and is constantly updated with true field measurements after each capacitor bank operation.

EOL meters are used to monitor the voltages at the lowest points on the feeder to ensure voltages do not drop below the minimum acceptable limits. The application has three user-defined EOL voltage levels — minimum, emergency and American National Standards Institute (ANSI) minimum. The minimum voltage represents the minimum desired voltage on EOL customers. As long as all EOL meters report voltages above this value, the system is said to operate as desired. If the voltage falls below this value, then the alarm is issued stating this occurrence.

The emergency voltage is used for ensuring the proper level for the user-defined minimum capacitor voltage. If the EOL voltage falls below the emergency level, the user-defined minimum voltage on the capacitor bank is raised and an alarm also is issued. The ANSI minimum voltage is set at 114 V and, if the voltage falls below this value, a major alarm is issued in addition to raising the user-defined minimum capacitor voltage.

The Algorithm

Switching commands are issued based on the following algorithm inputs — real (P) and reactive (Q) power at the head of the feeder, substation bus voltage, capacitor bank voltages and EOL voltages. Based on the substation bus voltage and capacitor bank voltages, the algorithm calculates the average voltage on the feeder and designates this value as the base value. In addition, the algorithm checks to see if EOL voltages are within the limits.

The algorithm simulates the feeder voltage profile for each of the capacitor banks in opposite status. This is done by using the voltage tables. For each of the possible combinations, the algorithm calculates the average voltage on the feeder as well as the power factor. The algorithm eliminates solutions that do not meet minimum or maximum capacitor bank voltages and solutions with a power factor outside the desired limits. User-defined voltage bandwidth is used to prevent hunting and to minimize the daily number of operations of capacitor banks.

Out of the remaining available solutions, the algorithm initiates the switching of the capacitor bank, which results in the lowest minimum average voltage. After each switching command has been issued, the algorithm waits 2 minutes to allow voltages to settle and records all inputs again. If all voltages are within the desired values, then the process is repeated until no switching commands have been issued. However, if any of the EOL voltages fall below the desired values, then the algorithm dynamically raises the minimum capacitor bank voltage as described previously.

The algorithm runs every 5 minutes, but during the time delay, two routines are run separately. The first routine dynamically changes the capacitor bank minimum voltage setpoint to ensure maximum energy reduction has been accomplished. After each switching has been completed, the algorithm compares the minimum average voltage as reported by EOL meters to the minimum capacitor bank voltage. If the minimum average EOL voltage is above the EOL minimum value setting by some value δ, then the algorithm changes the minimum capacitor banks voltage setting by lowering it down by some value less than δ. This is done to further maximize the energy demand reduction.

The volt/VAR control optimization results in a lower voltage profile than the existing voltage profile. In turn, this results in lower energy consumption and lower reactive requirements on the system.

The second routine run in a simulation mode determines the substation transformer optimum tap setting based on the ability of the algorithm to provide the flattest voltage along the feeder. BGE changes taps on the transformers twice a year based on the season loading profiles. Older transformers have 2.5% voltage change per tap, while newer transformers have 1.25% voltage change per tap.

The second routine assumes the initial voltage profile on the feeder to be with each of the lower taps. The goal of this routine is to provide the flattest-possible voltage profile along the feeder. If the transformer operates at a tap 4 and, if it is possible to obtain the flat voltage profile that satisfies all of BGE's constraints with a lower tap — for example, tap 3 — then the lower tap is the optimum tap. At lighter loads, it might be possible to achieve optimization with the tap at position 2. As a result, the algorithm provides an analysis of the optimum tap position.

New volt/VAR control with the objective of flattening the voltage profile across the feeder can result in different optimum tap positions. If the chosen optimum tap is a higher value, then during the days representative of the lower optimum tap position, to minimize energy demand, the algorithm will change the objective function to minimize the average voltage on the feeder rather than flatten the voltage. If the chosen optimum tap is a lower value, then an additional two-stage capacitor bank needs to be installed close to the substation to achieve the voltage profile during the days representative of the higher optimum tap position. (A 1,200-kVAR capacitor bank results in one LTC tap with ~0.85-V voltage raise on the substation bus; a single tap that changes voltage by 2.5% or 3 V would require approximately four 1,200-kVAR capacitor banks.)

For measurement and verification purposes of BGE's volt/VAR optimization, additional devices such as six GridSense LineIQ LT40 line load trackers for the line-loss measurements were installed along with nine GridSense TransformerIQs used for measuring the distribution transformer internal losses, voltage drop, and oil and winding temperature.

Next Step

BGE's smart grid distribution system pilot project was deployed in June 2012. After the initial data collection for volt/VAR control, an independent third party, Utilidata Inc., will analyze the data and provide the expected system cost-benefits to determine if a wider scale implementation is justified.

Paul J. Frey ( is manager of BGE's smart grid distribution system pilot project. He started with BGE as a test engineer and has held positions in gas and electric operations, as well as planning, strategic customer planning, asset management, system planning and equipment engineering. Frey received his BES degree from Johns Hopkins University and a master's degree in engineering administration from George Washington University.

Aleksandar Vukojevic ( joined BGE as a senior engineer in the smart grid distribution system pilot project department. During his career, Vukojevic has worked as a lead power systems engineer for smart grid technologies at GE, a field test engineer and transmission planning engineer at Georgia Power, and a smart grid engineer and system protection and controls engineer at BGE. He received a bachelor's degree in applied mathematics from Kennesaw State University, BSEE and MSEE degrees from the Georgia Institute of Technology, and a MBA degree from Robinson College of Business at Georgia State University. He is EIT certified.

Michael S. Smith ( is a lead engineer/work leader with BGE's automation and technology unit. Smith has 30 years experience in the utility business specializing in system protection, substation integration and supervisory control and data acquisition.

Companies mentioned:


Cooper Power Systems |

GridSense |

Landis+Gyr |

OSIsoft |



SUBNET Solutions |

Utilidata Inc. |