An industrywide demonstration shows voltage optimization could result in a 2% to 3% energy reduction or more.
Energy-efficiency policy in the electric power sector has often focused on consumers through end-use energy-conservation programs and incentives. However, significant opportunities also exist to reduce energy use through investments in the electric power supply infrastructure.
Besides reducing the losses in the infrastructure, there are also opportunities to reduce energy use on the customer side and peak demand through voltage management on the distribution system. A demonstration project has been designed to address the full range of these opportunities for energy savings. In many cases, the investments can be attractive alternatives to customer-side incentives and initiatives.
Potentials for Energy Savings
As part of an ongoing industry collaboration known as Green Circuits, 42 distribution circuits have been evaluated, with detailed simulations validated based on monitoring data, and six pilot demonstration projects have been operational for two to 12 months. Distribution losses range from less than 2% of energy delivered to almost 7%. The following various options for improving distribution efficiency and reducing losses were modeled and simulated as modifications to the base-case model:
Phase balancing — Rearranging loads on each phase of the circuit to reduce residual flows
Reactive power optimization — Additional capacitor banks or altered switching schemes
Selected reconductoring — Replacing selected conductor sections with larger, lower-resistance conductors
High-efficiency distribution transformers — Replacement of lower-efficiency line transformers with higher-efficiency units
Voltage optimization or conservation voltage reduction (CVR) — Intentional lowering of distribution circuit voltage within the lower band of the allowed American National Standards Institute range.
Of these, voltage optimization was the most consistently beneficial option. Improvements to reactive power profiles — normally through capacitor optimization — and phase balancing provided cost-effective improvements in about 10% to 20% of cases. Voltage optimization provided cost-effective improvements in at least 80% of cases.
Each circuit was modeled in detail from the substation to each customer meter. The analysis used a common platform, an open-source distribution system electrical simulation package known as OpenDSS, created by the Electric Power Research Institute (EPRI). Nearly all of the circuit models were augmented with historical circuit-measurement data that allowed for hourly simulation resolution of actual circuit-load patterns. This high-fidelity representation of the circuits' electrical characteristics, with the temporal and spatial diversity of the circuit loads, allows for capturing all loss sources. Several circuit models were also included:
Primary distribution lines and cables (three-phase mains and single-phase laterals)
Substation power transformers
All distribution service transformers
Secondaries and services (detailed models used when available)
Voltage-regulation controls (load tap-changing transformers, regulators and capacitors)
Detailed load representations, including the temporal variations in the load over the entire year.
Voltage optimization strategies were developed on a circuit-by-circuit basis. In general, the strategy was developed without significant investments in new voltage regulators and capacitor banks. Voltage reduction was achieved by changing strategies with existing technologies on the circuits. The standard voltage-reduction approach involved the following assumptions:
Use end-of-line feedback on all load tap-changing transformers and voltage regulators
Voltage set point = 118.5 V
Bandwidth = 2 V (+/- 1 V)
CVR factor for watts = 0.8
CVR factor for VARs = 3.0.
The median reduction in energy was 2%, with upper and lower quartiles of 2.8% and 1.3%. Because a constant CVR factor was used, these simulations mainly show how much room there is to drop voltage across the circuit throughout the course of an annual cycle. In about 10% of the circuits, the benefit was small or even negative. This is a result of limited availability to reduce the voltage on these circuits compared to the base case or cases where the existing voltage control was inadequate.
Voltage optimization can reduce energy consumption by several percent. Losses and end-use consumption are reduced when feeder voltage alone is managed to be within the lower end of the standard supply service voltage band. Additional reduction of both peak demand and energy consumption is possible with more aggressive investments in distribution voltage control to provide a flatter voltage profile. While end-point voltage monitoring technology is available, results of the initiative indicate it is also possible to obtain similar results using simple line drop compensator settings on these regulators.
EPRI's Green Circuits distribution project is validating the results of the Northwest Energy Efficiency Alliance (NEEA) Distribution Efficiency Initiative Study. Both indicate a 1% to 3% energy savings and a 1% to 4% demand reduction. A 5% to 10% reactive-power reduction can be achieved through voltage optimization. Assuming voltage optimization use achieves a nationwide penetration of 25% to 50%, an approximate annual savings range of 4 million MWh to 28 million MWh could be realized.
Validating with Field Results
Six circuits at four utilities are operating on a test program to evaluate reduced-voltage operating modes. The monitoring periods of the six circuits range from three to 12 months. Reduced voltage is evaluated by alternating daily between a normal-voltage mode and a reduced-voltage mode. Voltage is controlled with local control of voltage regulators or load tap-changer transformers. Most of the circuits are in the southeastern United States and most have mainly residential load.
The CVR factors, ranging from 0.7 to 0.9, are consistent with the NEEA study, possibly somewhat higher. The higher CVR factors are most likely due to the impact of having less electric heating loads in the Southeast compared to the Pacific Northwest. The confidence intervals are relatively tight, except for circuit F, which had only 64 days of monitoring because of some recorder problems. Down-line primary voltage reductions and customer voltage reductions are likely to be different.
Although most of the circuits in this field trial implement relatively simple voltage control, more advanced forms of voltage optimization are possible that could increase savings:
Voltage-measurement feedback to more precisely control voltage
Advanced customer metering to evaluate and control voltage based on meter voltage
More precise voltage control using capacitors and voltage regulators to flatten voltage profiles
Coordinated volt/VAR control to optimize voltages along with reactive power.
Some stimulus smart grid demonstrations will afford an opportunity to validate what additional gains, regarding voltage optimization, can be achieved through an advanced distribution automation and metering infrastructure. So while more technical work needs to be accomplished to better-predict benefits of voltage optimization and ensure unintended consequences do not outweigh the potential benefits, voltage optimization clearly looks like a promising technique. Cost data for the six pilot projects is in the process of being collected and analyzed. More detailed and utility-specific case results will be presented in the next article in this series on distribution efficiency and voltage optimization.
Need for Other Enablers
Demonstrations and technology are not the only limiting factors in improving the efficiency of the electric system. While there are many existing incentives for utilities to deploy end-use conservation programs, there are often too few incentives for them to invest in the distribution infrastructure to reduce energy use and peak demand. Revenues are usually linked to kilowatt-hour sales. Therefore, changes in incentive structures may be required for more broad deployment of utility-side options.
One of the important next steps to facilitate energy savings with voltage optimization is the development of a load model library that can be used with utility planning tools to accurately analyze different alternatives for controlling voltage and the resulting impacts. These models can be determined from the various deployments now under way. The benefits of voltage optimization are becoming quite clear, and now it is a matter of continuing to refine the models so benefits can be predicted and documented in a consistent manner. This will also help in justifying the required investments for regulators and documenting the energy-savings potential for overall energy-conservation portfolios.
Industry demonstrations such as the Distribution Efficiency Initiative and the Green Circuits collaboration are intended to increase awareness, identify technology options and business transformation requirements, develop comprehensive measurement and verification protocols, and validate the potential savings, which might lead to valuation of efficiency improvements in power delivery. Using the distribution system as an energy-reduction and peak demand resource may end up being more cost effective, controllable and sustainable than other end-use efficiency options.
Karen Forsten (email@example.com) is the area manager responsible for EPRI's research and development related to T&D efficiency, renewable integration, operations and planning. She manages a team of 40 technical staff and engineers and more than US$16 million in related research, development, demonstrations, applications and services supporting the power delivery and utilization sector at EPRI.
Tom Short (firstname.lastname@example.org) is part of EPRI's distribution systems group and has more than 20 years experience in distribution system studies, reliability, power quality, distributed generation, lightning protection and capacitor application. He authored the Electric Power Distribution Handbook and two spin-off books based on the handbook: Distribution Reliability and Power Quality and Electric Power Distribution Equipment and Systems. He is currently working on research in the areas of distribution system loss reduction, fault location and arc flash.
Ronald C. Belvin (email@example.com) is director of distribution planning for Duke Energy in the Carolinas. He has 29 years experience in distribution management and engineering. He is currently serving as chair for EPRI's Distribution Efficiency Program 172 and Green Circuits. He has a BSEE degree from North Carolina State University, a MBA degree from Queens University of Charlotte and is a professional engineer.
K.C. Fagen (firstname.lastname@example.org) is a senior project manager for R.W. Beck Inc., an SAIC company. With 19 years experience in distribution system planning, design and control systems, he was the project manager for the distribution efficiency initiative and is supporting EPRI on similar projects. In addition, Fagen is helping Bonneville Power Administration expand its distribution efficiency program including voltage optimization. Fagen is a professional engineer.
|Circuit||Voltage||Number of customers||Circuit miles||U.S. location||Percent residential||Voltage-control devices|
|A||34.5 kV||3597||105||Southeast||75||Station LTC|
|B||12.5 kV||1545||73||Southeast||97||Station and two line regulator sites|
|C||12.5 kV||1379||48||Southeast||96||Station and one line regulator site|
|D||23.9 kV||2867||48||Southeast||94||Station LTC|
|E||23.9 kV||2088||37||Southeast||94||Station LTC|
|F||4.2 kV||501||3||Northeast||70||Station regulators|
Table 1 provides characteristics of the six test circuits. Most of them are in the southeast United States and most have mainly residential load.
|Circuit||Monitor days||Linear model R2||Average voltage reduction||Energy reduction (95% confidence interval)||CVR factor|
|A||362||0.953||3.62%||2.71% (2.20, 3.22)||0.75|
|B||289||0.984||2.81%||2.47% (2.01, 2.93)||0.88|
|C||345||0.989||3.57%||2.38% (1.93, 2.82)||0.66|
|D||153||0.981||1.89%||1.64% (1.24, 2.04)||0.87|
|E||153||0.993||1.89%||1.73% (1.27, 2.19)||0.92|
|F||64||0.937||2.76%||1.82% (0.07, 3.57)||0.66|
Table 2 shows the results from the normalized regression model used to estimate the energy reduced along with the confidence interval.