Several utilities are piloting field trials aimed at improving feeder efficiency, among them are Duke Energy and Alabama Power Co. (APC). The utilities are taking specific energy-conservation measures, including system improvements, to reduce losses and operation in the lower half of the American National Standards Institute (ANSI) C84.1-2006 voltage range A. The preliminary results of these pilots are confirming a potential energy reduction of 1.6% to 2.7%.

Study Approach

The Distribution Green Circuits research collaborative program of the Electric Power Research Institute (EPRI) was designed to use analytical models to verify circuit performance after establishing a baseline for the study. Of ultimate interest is the determination of energy-savings potential. System improvements to increase the operating efficiency (i.e., reduce losses or flatten the voltage profile) were evaluated and new voltage settings were determined. Based on hourly load data, the models estimated the reduction in losses and energy-consumption savings resulting from the system improvements.

Next, field trials were used to evaluate the response of load to a voltage reduction. This is quantified with the conservation voltage reduction (CVR) factor (CVRf = %ΔE/%ΔV), which is the percentage reduction in energy consumption for a 1% voltage reduction. In the field trials, demand, energy and reactive power impacts were measured using a day-on/day-off alternating scheme, changing control settings from normal to reduced on alternate days. After enough operating time, the variations due to conditions outside of the control parameters, such as weather, are less significant, allowing the causes and effects of voltage optimization to be determined with a higher degree of confidence.

Because the energy-consumption changes were on the order of 1% to 3% and loading is highly variable (with the ability to vary 30% to 50% during a 24-hour period), it can take many months or years of data sampling to be confident about the results. Using a regression model to normalize the voltage effect, including other variables, helps reduce the amount of data needed to get a statistically significant estimate of the voltage effect. EPRI used a regression model with a comparable circuit to normalize for load variations.

Duke Energy Circuits

Two of the Duke Energy circuits in the pilot are characterized as mainly residential with a high load density: 59 and 55 customers per primary circuit mile of primary conductor (approximately 35 customers/km). The two circuits, referred to as circuits 1 and 2, were the only loads served by their substation buses and regulated by the substation load tap changer (LTC). The circuits operated at 24 kV and were relatively short in length, making them good candidates for operating at the reduced voltage levels.

To minimize the effects of voltage drop, both circuits were evaluated using OpenDSS, EPRI's opensource software-modeling tool. These circuits already had a fairly flat voltage profile and low primary line losses. They did not require additional equipment or system improvements to improve the voltage drop, power factor or phase balancing.

Computer simulations predicted energy savings of between 1.5% and 2.5%, and a reduction of losses between 2.4 kW and 7.0 kW associated with the voltage optimization. The loss savings were on the utility side of the meter and included the primary conductors and distribution transformers.

Based on the results from the simulation, line drop compensation (LDC) was chosen to lower the average voltage level, and Duke Energy implemented the LDC settings. Duke updated its regulator controls for the substation LTC to include supervisory control and data acquisition (SCADA) control logic.

For both circuits, the LTC controller was set with an R setting but not the X setting. This made the controller insensitive to reactive power flows resulting from capacitor switching. Duke alternated the LTC controls between normal and reduced voltage modes using an algorithm developed in house. Both Duke circuits had advanced metering infrastructure (AMI) information available at residential customer meters, helping to identify areas with low voltage.

With the new LTC settings, Duke Energy demonstrated a real-life savings consisting of 1.64% and 1.73% energy reductions at a 95% confidence level. This resulted in a CVRf of 0.87 and 0.92, and aligned with the model-simulation results very well.

Based on a typical generation mix for Duke Energy, implementing voltage optimization on just these two circuits reduced its carbon-dioxide output by about 1700 tons per year. The cost associated with implementing this program was minimal at US$25,000 per circuit. The benefit-cost ratios for both circuits exceeded 4.

Alabama Power Co. Circuits

APC joined EPRI's Green Circuits collaborative to gain a better understanding of how voltage optimization could be used to reduce energy requirements. Georgia Power Co., a sister company to APC, has had a successful voltage-reduction program in place for the past decade, which has been used as a peak shaving demand-response program during critical periods. Two APC circuits, referred to as circuits 3 and 4, were included in field trials. Circuit 3 was a primarily urban residential circuit, while circuit 4 primarily served rural residential load. Circuit 3 was regulated by the substation LTC, and circuit 4 had dedicated per-phase feeder regulators at the substation and three sets of line regulators out on the feeder, two of which were in series on one fork of the feeder. As with Duke's circuits 1 and 2, EPRI's OpenDSS was used to evaluate APC's circuits 3 and 4.

For circuit 3, APC installed switched capacitor banks and per-phase volts-ampere-reactive (VAR) control with voltage override. Crews also performed minor phase balancing to correct some small unbalances. Circuit 4 originally had five fixed capacitor banks and one switched bank. With this configuration, the power factor at peak load was good, but at lighter loads, the power factor was leading, causing extra line losses and higher circuit voltages. The capacitors were reconfigured using the same per-phase technique and settings as on circuit 3. Moderate phase balancing work was done to strike a balance between summer and winter unbalance conditions.