Historically, traditional voltage regulation techniques based on power-flow modeling have been sufficient for distribution system control: delivery of a nominal service voltage with a range of +/- 5% (e.g. 114V-126V) to an unengaged electric energy consumer with one-way power flow. However, the continuing modernization and automation of the electric grid brought about by the introduction of Smart Grid technologies require more advanced management and control than what has been available to utility operators in the past. One of the operational goals that utilities tried to achieve using traditional techniques is Conservation Voltage Reduction (CVR).
CVR is the operation of the electric distribution system in a manner which delivers voltages to consumers at or near the lower bounds of a utility’s delivery standards (and nearer the nameplate voltage of consumer devices) in order to achieve peak demand reduction, energy savings, and reduction in technical and non-technical line losses.
Since 1970, CVR has been proven by many utilities to reduce energy consumption and lower demand by 1 to 3 percent. Typically, these savings have been achieved through the application of complex power-flow models and Line Drop Compensation technology. However, this approach is not well suited to the modern grid (e.g. distributed energy resources, etc.). New technologies have recently been introduced to the industry that allow utilities to realize the benefits of CVR without the constraints of traditional approaches. One of these new approaches leverages signal processing and adaptive control technology in an innovative way to implement intelligent voltage control.
Fueled in part by the American Recovery and Reinvestment Act of 2009, CVR has become an operational goal of a technology referred to as Volt/VAR Optimization or VVO (also known as Integrated Volt/VAR Control or IVVC). In addition to the historical energy benefits achieved through voltage reduction, modern VVO technologies also improve distribution system reliability and alleviate the de-stabilizing impacts of distributed energy resources (e.g. photovoltaics).
On suitable electric distribution systems, an application of CVR may achieve peak demand reduction and energy savings of up to 5 percent. The business case of CVR as an energy efficiency tool is significantly affected by the amount of voltage reduction that any particular technology can achieve through its control algorithms. A fifty percent increase in voltage reduction (e.g. an increase from two percent to three percent) can more than double the net present value of the business case.
However, traditional power-flow based methods of CVR –
- Simple Voltage Reduction that uses Automatic Voltage Regulator Controls (AVR) to control peak-load voltage set points
- Line Drop Compensation that measures the voltage drop along the length of the feeder
- Distribution Management Systems (DMS), which use real-time switching information – while capable of achieving nominal levels of voltage reduction, have often done so at the expense of increased voltage regulator (i.e. on-load tap changing transformer or auto-regulating voltage transformer) tap change operations.
Transmission v. Distribution: Power-flow modeling is a powerful tool for system planning and load forecasting, allowing the complexities of the grid to be simplified in such a manner as to be manageable by a utility’s distribution planning team and its cost effective computational tools.
Along with significant advances in computing power (and associated reduction in cost) however, manageable models have become more complex. What began as a simple model of non-time varying, linear circuit elements, has advanced to include non-linear elements and time-varying loads. The final piece of the solution puzzle of the power-flow modeling science is to include loads that are randomly allocated in both space and time. The resultant power-flow model is now written as a stochastic differential equation. This approach has achieved relative success in transmission management systems; however, these same models begin to suffer from burdensome intricacy and often become non-convergent when addressing the electric distribution system and its thousands of delivery nodes and increasing number of load inputs (e.g. distributed generation).
Regardless of how advanced power-flow models and computing resources may become, the approach outlined above will require that all circuit elements be known with precision (which for many utilities is not cost effective). Any change to customer demographics, load types, circuit switching or grid infrastructure will impact the accuracy of power-flow model results. On bulk power transmission systems, inaccuracies (of which there are relatively few) rarely impact the results of power-flow based control. However, on a distribution system where voltage and VARs are managed to a much narrower range (e.g. 2-3%) in order to gain efficiencies, these inaccuracies will lead to lost benefits and may impact system reliability.
Set-Point Dispatching: The majority of the advanced DMS solutions issue voltage set-points to field controllers based on the results of their power-flow calculations. These local devices then control their elements based on local measurements – not the circuit-wide measurements of voltage or VARs – leading to non-optimal decisions for management of the entire circuit. The consequence of this type of dispatch is that local field controllers continue to operate in an uncoordinated fashion, leading to hunting and dithering between voltage regulator elements; and, control instability arises when voltage regulators try to corrected voltage rise and drop caused by capacitor switching.
Traditional Voltage Tap Changer Control: Digital voltage regulator tap changer controls issue a tap change under normal conditions using three parameters: a local voltage set-point (target or voltage level), a control bandwidth and a computation timer.
This simple approach is merely a reactive decision to voltages that are influenced by random consumer behavior, and suffers from a fundamental problem: the tap changer will only make a decision when local voltages cross control bandwidth boundaries. The outcomes of this reactive decision are two-fold:
- The tap changer will not bring the voltage back to the optimal target, even if the voltage has spent significant time near the boundaries
- The tap changer may make unnecessary decisions for momentary voltage excursions beyond the boundaries.
As such, maximization of voltage reduction is directly linked to the frequency of tap change operations.
De-tuning the System: In order to reduce the increased number of tap change operations caused by set-point dispatching and traditional voltage control, several parameters are adjusted within the DMS or local control elements (specifically OLTC and voltage regulator controllers), effectively ‘de-tuning’ the optimization goals of the VVO solution. There are several primary de-tuning methods utilized for set-point dispatching of a VVO system, including:
- Decreasing the dispatch frequency of set-points
- Increasing voltage bandwidth
- Increasing the length of the computational timer.
Issuance of a new voltage set-points may cause an immediate tap change in a voltage regulator; therefore, many CVR/VVO solutions have reduced the frequency with which they issue voltage set-points. Voltage tap change frequency can also be reduced by increasing the voltage control bandwidth, or dead band. Increasing the length of the computation timer is typically used when system instability causes extraneous tap changes (usually due to capacitor switching or voltage rise caused by distributed generation). However, all of these methods are ineffective for optimizing voltage control to achieve energy and demand savings, as they decrease the amount of time spent at the optimal voltage level.
Challenges Presented by the Modern Grid
Optimization of the modern grid requires more advanced management and automated control than has been available to utilities in the past. One of the primary focuses of grid modernization is to enable the integration of customer-owned distribution generation resources such as photovoltaics and wind power. There are many significant issues created by these new generation sources that can impact grid stability:
- Random behavior caused by supply intermittency
- Multi-direction power flow
- Voltage spikes and sags
- Extended periods of high voltage
- Increased tap change operations
- Increased capacitor switching.
The combination of the complexities grid modernization and the limitations of traditional power-flow based, voltage control regimes in achieving optimal voltage levels without spiking asset operations and threatening reliability, have brought about the need for advanced, real-time solutions for intelligent voltage control.
As an alternative to power-flow modeling, which treats voltage fluctuation as an impact on circuit conditions, voltage fluctuations can be looked at as a process problem caused by demand, which is driven by stochastic (i.e. random) consumer behavior. This stochastic behavior, when viewed as a 24-hour demand profile, follows a pattern that is obvious to the human observer and utility load forecasters. However, from minute to minute, the demand behavior exhibited by consumers is random. The impacts of this behavior can be observed in real-time via the voltage signals created by the demand process. In addition to consumer (i.e.) load behavior and its influence on voltage, there are two other broad categories of behavior that influence voltage measurement on the grid:
- The behavior of grid inputs (i.e. transmission sources, generation resources, distributed energy resources, etc.)
- The behavior of losses on the grid structure
Over time, the impacts of these behaviors are reflected in the voltage signals measured on the distribution system. One example of this is the change in system voltage as load increases and decreases around the daily system peak.
Intelligent voltage optimization that leverages signal processing delivers more demonstrated benefits by way of its observation and extraction of behavioral information from voltage signals along an entire distribution circuit. Rather than making a decision based only on the voltage measured at a single control element, the resulting adaptive control decisions are made and executed based on that actionable intelligence and are not based on circuit modeling.
Elasticity: As previously stated, the main issue with the legacy approach to voltage regulation is that the AVR only makes a decision when voltage crosses a boundary threshold, turning the control decision for a stochastic process into a binary reaction. In order to more appropriately implement control of this process, control algorithms should instead use statistical control, allowing the voltage controller to make a decision based on the probability of the need for a tap change, rather than to react to a voltage change.
Combined with the behavioral information gleaned from the application of signal processing, this control algorithm provides several benefits over traditional methods, the first of which is that the VVO system makes only the decisions necessary to achieve the optimal voltage target. For example, it can identify momentary voltage excursions beyond the boundary threshold, perhaps caused by distributed generation (DG), and prevents reactions that would need to be immediately corrected after the voltage inevitably returns to within the control bandwidth. Similarly, the algorithm can identify when the voltage is beginning to trend outside of the bandwidth and can make a tap decision to maintain the voltage near the target without the voltage ever crossing the boundary threshold.
This process-based approach to implementing intelligent voltage control combines smarter and less frequent tap operations with more time spent near optimal voltage levels. It is innovative VVO that not only achieves voltage reduction, but also helps to improve distribution system reliability and to alleviate the de-stabilizing impacts of distributed energy resources. As the grid continues to modernize and automate, more such advanced approaches will be necessary.
About the Author
Jeremy Wilson started with Programmable Control Services, Inc. in 2003 as a Sales and Marketing Coordinator. Wilson quickly advanced through the company, holding a variety of positions. In 2012, he was on the founding team of Utilidata, Inc., where he serves as Director of Technical Sales. In this role, Wilson is responsible for developing partner relationships and alliance agreements with large utility clients and other smart grid vendors; working in partnership with regional energy authorities to advance regulatory framework for the implementation of Volt/VAR Optimization; and, product management of Utilidata’s digital technology. Jeremy also makes frequent presentations at trade shows and educational conferences. Wilson graduated Magna Cum Laude with a BA from Washington State University.
