A Scalable Solution for Grid Optimization and DER Integration Cooperative Research and Development Final Report CRADA Number: CRD-20-16909 NREL Technical Contact: Killian McKenna NREL is a national laboratory of the U.S. Department of Energy Technical Report Office of Energy Efficiency & Renewable Energy NREL/TP-6A40-94043 Operated under Contract No. DE-AC36-08GO28308 July 2025 This report is available at no cost from NREL at www.nrel.gov/publications. A Scalable Solution for Grid Optimization and DER Integration Cooperative Research and Development Final Report CRADA Number: CRD-20-16909 NREL Technical Contact: Killian McKenna Suggested Citation McKenna, Killian. 2025. A Scalable Solution for Grid Optimization and DER Integration: Cooperative Research and Development Final Report, CRADA Number CRD-20-16909. Golden, CO: National Renewable Energy Laboratory. NREL/TP-6A40-94043. https://www.nrel.gov/docs/fy25osti/94043.pdf. NREL is a national laboratory of the U.S. Department of Energy Technical Report Office of Energy Efficiency & Renewable Energy NREL/TP-6A40-94043 Operated under Contract No. DE-AC36-08GO28308 July 2025 This report is available at no cost from 15013 Denver West Parkway NREL at www.nrel.gov/publications. Golden, CO 80401 303-275-3000 • www.nrel.gov This work was authored by NREL for the U.S. Department of Energy (DOE), operated under Contract No. DE- AC36-08GO28308. Funding provided by the U.S. Department of Energy Office of Energy Efficiency and Renewable Energy Solar Energy Technologies. The views expressed herein do not necessarily represent the views of the DOE or the U.S. Government. This work was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, nor any of their contractors, subcontractors or their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or any third party’s use or the results of such use of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof or its contractors or subcontractors. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof, its contractors or subcontractors. U.S. Department of Energy (DOE) reports produced after 1991 and a growing number of pre-1991 documents are available free via www.OSTI.gov. Cover photos (clockwise from left): Josh Bauer, NREL 61725; Visualization from the NREL Insight Center; Getty-181828180; Agata Bogucka, NREL 91683; Dennis Schroeder, NREL 51331; Werner Slocum, NREL 67842. NREL prints on paper that contains recycled content. Cooperative Research and Development Final Report Report Date: March 26, 2025 In accordance with requirements set forth in the terms of the CRADA agreement, this document is the CRADA final report, including a list of subject inventions, to be forwarded to the DOE Office of Scientific and Technical Information as part of the commitment to the public to demonstrate results of federally funded research. Parties to the Agreement: Utilidata, Inc. CRADA Number: CRD-20-16909 (Project 1 of 2) CRADA Title: A Scalable Solution for Grid Optimization and DER Integration Responsible Technical Contact at NREL: Killian McKenna | killian.mckenna@nrel.gov (Lead PI) Co-Authors: • Kapil Duwadi • Francisco Flores-Espino • Aadil Latif • Rasel Mahmud • Subhankar Ganguly • Jing Wang Name and Email Address of POC at Company: Marissa Hummon | mhummon@utilidata.com Sponsoring DOE Program Office: Office of Energy Efficiency and Renewable Energy (EERE), Solar Energy Technologies Office Joint Work Statement Funding Table showing DOE commitment: No NREL shared resources. *Estimated NREL Shared Resources Costs (aka Government In-Kind) Year 1 $0.00 Year 2 $0.00 Year 3 $0.00 Total $0.00 * Only for Project 1 (Original Agreement and Modification 2) under this CRADA number. Project 2 (Modification 1) is unrelated, and its final report was published separately. 1 Executive Summary of CRADA Work: Utilidata Inc. is an energy software company that provides clean energy and digital solutions for utilities. This work was the result of Utilidata Inc. being awarded a New York State Energy Research and Development Agency award to integrate Utilidata volt-VAR optimization (VVO) software with smart inverters to explore enhanced voltage control. This project was conducted in partnership with Utilidata Inc., National Grid, the New York State Energy and Research Development Authority, and the National Renewable Energy Laboratory (NREL). As part of the project NREL conducted modeling and simulation analysis, developed cost-benefit analysis framework, and conducted laboratory testing to understand the potential for smart inverter integration to a centralized VVO control scheme. The remainder of the report provides an executive summary of the work, an introduction to the challenges, sections detailing specific aspects of the work, and the overall findings and conclusions. CRADA benefit to DOE, participant, and U.S. taxpayer: • Uses the laboratory’s core competencies and/or • Enhances U.S. competitiveness by using DOE developed intellectual property and/or capabilities. Summary of Research Results: Note: This report includes only Project 1 (Original Agreement and Modification 2) under this CRADA number. Project 2 (Modification 1) is unrelated, and its final report was published separately. Task 1- Perform techno-economic system study: NREL developed distribution feeder models to create a simulation environment for time-series power-flow evaluation of volt-VAR optimization (VVO) and smart-inverter (SI) strategies. Additional details for all subtasks are provided in Section 0 of this report. Subtask 1.1 Feeder Conversion and Validation for Steady-State Simulation: NREL converted four distribution feeders fed by one substation from National Grid geographical information system (GIS) and Synergi power-flow models to OpenDSS, using NREL’s Distribution Transformer Tool (DiTTo) conversion. This established the power-flow model for steady-state simulation. Subtask 1.2 Create Black-Box Co-Simulation and Time-series Simulation: NREL created an OpenDSS quasi-static time-series co-simulation framework for power-flow, leveraging National Grid advanced metering infrastructure (AMI), supervisory control and data acquisition (SCADA), DER technical information and forecasts, and voltage control setting datasets. This enabled dispatch by Utilidata’s VVO for time-series powerflow. Subtask 1.3 Conduct Scenario Modeling of VVO and SI: NREL conducted multiple scenarios with and without VVO and SI controls to assess benefits and impacts, specifically examining change in energy consumption, PV curtailment, PV integration, energy imports, and device lifetime (i.e., potential increased wear of voltage control assets). 2 Subtask 1.4 Estimate Costs of Scenarios: NREL developed multiple metrics (e.g., energy reduction from changes in supply voltage, PV curtailment, substation energy imports, etc.) that could be used in later estimates of benefits and costs of VVO and SI integration and control. Task 2 – Develop benefit-cost analysis (BCA) framework: NREL developed a BCA framework for assessing the benefits of VVO, the analysis is detailed in Section 0 of this report. Subtask 2.1 Conduct Preliminary BCA: NREL established a BCA framework and obtained feedback from Utilidata, National Grid, NYSERDA, and DER developers. Subtask 2.2 Conduct Final CBA: NREL presented final and refined BCA framework presented and benefits calculated from energy reduction and curtailment that were simulated in task 1. Task 3 – Conduct laboratory testing: NREL established a power hardware-in-the-loop (PHIL) test set-up and evaluated VVO SI communication strategies against two 250 kW solar inverters similar to those being used in a solar ground-mount installation connected to National Grid’s distribution feeders. The test set-up and results are detailed in Section 3 of this report. Subtask 3.1 Hardware Laboratory Set-up: NREL established OPAL-RT ePhasorSim real- time models of National Grid distribution feeders, and configured SI communication set-up for external controllers communication through DNP3 and RS-485 protocols, and a grid simulator, to the two smart-inverters. Subtask 3.2 Software Laboratory Set-up: NREL set-up the ability for VVO to dispatch voltage control devices in the real-time model for running detailed co-simulation studies. Subtask 3.3 Conduct VVO and SI PHIL Tests: NREL conducted SI tests and used time-series VVO control from task 2, with a detailed focus on communication with SI to examine open-loop and closed-loop communication lags and distributed autonomous control. Subtask 3.4 Review and Provide Feedback to CYME and ASPEN SI Models: This task was removed in agreement with the partner. Task 4 - Field Deployment Support: This task was removed in agreement with the partner. Task 5 – Project Management, Stakeholder Engagement and Reporting: NREL established regular cadence of feedback and presented results for tasks 1 through 3 through engagement with National Grid, Utilidata, NYSERDA, and smart inverter developers. 3 Project Background and Summary of Work: Utilidata Inc. is an energy software company that provides clean energy and digital solutions for utilities. This work was the result of Utilidata Inc. being awarded a New York State Energy Research and Development Agency award to integrate Utilidata volt-VAR optimization (VVO) software with smart inverters to explore enhanced voltage control. This project was conducted in partnership with Utilidata Inc., National Grid, the New York State Energy and Research Development Authority, and the National Renewable Energy Laboratory (NREL). As part of the project NREL conducted modeling and simulation analysis, developed cost-benefit analysis framework, and conducted laboratory testing to understand the potential for smart inverter integration to a centralized VVO control scheme. The remainder of the report provides an executive summary of the work, an introduction to the challenges, sections detailing specific aspects of the work, and the overall findings and conclusions. 4 List of Abbreviations and Acronyms AMI advanced metering infrastructure BCA benefit-cost analysis CRADA cooperative research and development agreement CVR conservation voltage reduction DER distributed energy resource DOE U.S. Department of Energy DTM Distribution Test Manager (Triangle MicroWorks) DuoS distribution-use-of-system EPRI Electric Power Research Institute EUT equipment under test IEEE Institute of Electrical and Electronics Engineers LTC load tap changer MW megawatts NREL National Renewable Energy Laboratory PHIL power hardware-in-the-loop POC point of contact PV photovoltaics SCADA supervisory control and data acquisition SI smart inverter VAR volt-amps reactive VO volt-VAR optimization 5 Executive Summary Utility distribution voltage control is fundamental to delivering power system stability and ensuring the high-level of power quality required by end-user equipment. Historically, the challenge for distribution utilities was dealing with voltage drop from the distribution substation (i.e., where transmission voltages [e.g., 69 kV and greater] are stepped down to distribution primary voltages [e.g., with typical voltages between 4 and –35 kV]) voltages to the customer. As power flows along utility assets (e.g., transformers and conductors), voltage drops due to the impedance and associated losses of these assets. Typically, utilities try to keep per unit voltages (e.g., 1 p.u. is 100% of nominal) high at distribution substation transformers in order to account for voltage drop as power flows down to the end user and meter. In the United States, it is typical for the substation to have on-load tap-changing capabilities (i.e., transformers that can modify their turns ratio, or tap position while on-load to produce voltages slightly higher or lower than 1 p.u.). Utility voltage control assets downstream of the substation (e.g., capacitor banks and line voltage regulators) can provide enhanced control capability and coordination opportunities, and they can allow utilities to operate voltages on the lower side of established bounds to achieve energy efficiency. Trying to enhance voltage control by coordinating voltage control assets to achieve some objective in terms of power quality and energy efficiency is referred to as volt- VAR optimization (VVO), where VAR is volt-amps reactive. And utilities can try to achieve multiple objectives with respect to VVO, including energy efficiency, power quality, and—of interest to this study—solar integration. VVO has the potential to reduce energy consumption by reducing voltages within utility- specified character-of-service bounds. As voltage decreases, end-use loads consume less energy. VVO also has the potential to increase hosting capacity by reducing the headroom for overvoltages, as overvoltages are typically the first problem encountered with high levels of solar integration. This was the focus of this project, funded by the New York State Energy and Research Development Authority, Utilidata Inc. partnered with the National Renewable Energy Laboratory (NREL), National Grid, and Standard Solar to advance VVO and smart inverter (SI) integration. Pre-Existing Low Pre-Existing High Customer Customer Voltages Pre-VVO Voltages Pre-VVO Few Voltage Low opportunity to Medium opportunity Control Assets reduce voltages/ to reduce voltages/ (e.g. just LTC) energy consumption energy consumption VVO Opportunity and increase hosting and increase hosting capacity capacity Low opportunity to High opportunity to Many Voltage reduce voltages/ reduce voltages/ Control Assets energy consumption energy consumption (e.g. regulators, and increase hosting and increase hosting capacity capacity capacitor banks) Figure ES-1. Opportunity for VVO depending on original network voltages and the number of voltage control assets LTC = load tap changer 6 The potential to achieve, or improve, energy efficiency and hosting capacity depends on the initial network conditions. Networks with existing low voltages will have low potential to reduce voltages; indeed, voltages may have to be increased, reversing the VVO benefits in favor of improving power quality and bringing voltages into character-of-service ranges. Networks with high voltages have higher opportunity for achieving energy reduction and increasing hosting capacity. The number of control assets (e.g., regulators and capacitor banks) also improves the potential opportunity for VVO. Figure ES-1 conceptually presents the conditions where VVO will have a greater potential to reduce voltages and energy consumption and to increase hosting capacity. Networks with existing low voltage conditions may also benefit by bringing voltages back into bounds while still ensuring maximum energy reduction and hosting capacity benefits. Of interest in this research was how higher levels of solar photovoltaics might impact traditional VVO benefits. Another key element of the project research was examining the potential benefits of integrating SI dispatch (i.e., active and reactive power dispatch) into a centralized VVO scheme. The benefits of SI integration to VVO were examined throughout the project. Three principal tasks were completed as part of the project: 1. Development of a simulation model to assess the power flow benefits of VVO 2. Assessment of how to approach cost-benefit analysis of VVO and SI integration 3. Laboratory demonstration and analysis of centralized versus distribution control for SI volt-VAR control. Overall, the key findings of this project were: • VVO can provide energy efficiency benefits by reducing voltages within character-of- service ranges and reducing energy consumption and losses. The benefits identified in this report that are specific to the feeders analyzed ranged from 1% to 2%. • VVO can help mitigate solar integration voltage control challenges by mitigating overvoltages. By mitigating voltage rises caused by solar, energy efficiency benefits can increase under VVO. • SI integration benefits into VVO heavily depend on the degree to which utilities have already deployed traditional voltage control assets (e.g., load tap changers, capacitor banks, and line voltage regulators). For the feeders analyzed, large-scale deployment and use of these assets enhance VVO benefits but leave little room for additional SI voltage control benefits. • Smart inverters can be integrated into a centralized voltage control scheme through the proposed communication pathways (detailed in Section 3). Centralized control was demonstrated through external remote control and communication for dispatching smart inverters in a laboratory setup at NREL’s Energy System Integration Facility. • When contrasting centralized control and distributed autonomous control by the inverter, communication lags from central dispatch and different ramp rates and response characteristics (linear versus first-order for centralized and autonomous respectively) were observed. 7 Table of Contents Executive Summary ..................................................................................................................................... 6 1 Introduction .......................................................................................................................................... 11 2 Modeling and Cost-Benefit Analysis of VVO and SI Dispatch ....................................................... 16 2.1 What should be considered in a benefit-cost analysis framework for VVO? .............................. 18 2.2 What is the value of VVO of the examined feeders?................................................................... 23 2.3 How might increased deployment of PV systems change the value of VVO? ............................ 25 2.4 How will advanced inverter functions from smart inverters impact the value of VVO? ............ 25 2.5 How can AMI enhance the value of VVO? ................................................................................. 26 2.6 Can VVO increase the amount of solar that distribution systems can accommodate? How can advanced inverter functions impact solar penetration? ................................................................ 27 2.7 Considerations for Implementing a BCA for Utilities ................................................................. 31 3 Laboratory Integration of SI Communication ................................................................................... 32 3.1 Overview of Testing for Remote Open-Loop, Remote Closed-Loop Control, and Distributed Autonomous Control .................................................................................................................... 32 3.2 Test Results .................................................................................................................................. 36 3.2.1 Remote Open-Loop Dispatch ......................................................................................... 36 3.2.2 Remote Closed-Loop Dispatch ....................................................................................... 37 3.2.3 Remote Closed-Loop Compared to Distributed Autonomous Inverter Control ............. 37 3.3 Overview of Power Hardware-in-the-Loop Testing .................................................................... 39 4 Conclusion ........................................................................................................................................... 43 5 References ........................................................................................................................................... 44 8 List of Figures Figure ES-1. Opportunity for VVO depending on original network voltages and the number of voltage control assets ............................................................................................................................. 6 Figure 1. Traditional and future voltage monitoring and control assets considered in this report. ............. 12 Figure 2. Constant impedance (Z), constant current (I), and constant power (P) active power and current responses with respect to voltage............................................................................................ 13 Figure 3. Information graphic on how original network conditions can determine the type of benefits VVO can bring ........................................................................................................................ 14 Figure 4. Schematic of a feeder using VVO control and the voltage assets being controlled..................... 17 Figure 5. Opportunity for VVO depending on original network voltages and the number of voltage control assets ........................................................................................................................... 18 Figure 6. Diagram of inputs to the BCA calculation ................................................................................... 19 Figure 7. BCA framework for VVO with key metrics from the perspectives of customers, solar owners, and utilities .............................................................................................................................. 20 Figure 8. Losses, PV production, and customer load components of the change in substation imports in the Base Case with and without VVO in 2019 ....................................................................... 24 Figure 9. Percentage of customer voltage hours in different voltage bands among various scenarios ....... 28 Figure 10. Percentage of customer voltage hours for the difference between actual voltages and 1.05 p.u., the high voltage limit, among various scenarios..................................................................... 29 Figure 11. Voltages at 5-MW non-rooftop site connected to primary network in the hosting capacity case study, examining voltages with and without VVO ................................................................. 30 Figure 12. Setup for open-loop and closed-loop centralized SI dispatch response time characterization .. 33 Figure 13. Overview of communication elements involved in centralized dispatch ................................... 35 Figure 14. Overview of time-series elements in communication elements in remote dispatch and tracking error ......................................................................................................................................... 35 Figure 15. Open-loop tests and corresponding inverter ramp rates for (a) a positive ramp rate and (b) a negative ramp rate ................................................................................................................... 36 Figure 16. Closed-loop tests and corresponding inverter ramp rates for (a) a positive ramp rate and (b) a negative ramp rate ................................................................................................................... 37 Figure 17. Distributed autonomous volt-VAR control (i.e., inverter-programmed volt-VAR control) responses to step changes in grid-simulator voltage ............................................................... 38 Figure 18. Remote closed-loop volt-VAR inverter dispatch (i.e., Triangle DTM inverter dispatch) responses to step changes in grid-simulator voltage ............................................................... 38 Figure 19. Setup for PHIL tests ................................................................................................................... 40 Figure 20. Inverter active and reactive power and three-phase voltages for (a) unity power factor, (b) distributed autonomous inverter volt-VAR control, and (c) remote closed-loop dispatch control for a PHIL test demonstrating the setup depicted in Figure 19 .................................. 42 9 List of Tables Table 1. Description of VVO BCA Metrics and Their Expected Cost and Impact..................................... 20 Table 2. Capacity of Rooftop and Non-Rooftop PV for Different Scenarios.............................................. 23 Table 3. VVO Resulting in a 1%–2% Reduction in Cost of Energy Imports in 2019 ................................ 24 Table 4. Volume of Customer Net Exports in 2035 in Annual Megawatt-Hours ....................................... 25 Table 5. VVO Cost Savings for 2035 .......................................................................................................... 25 Table 6. Key Tests and Findings Based on Testing Goals .......................................................................... 33 Table 7. Estimated Communication Lag Time and Positive and Negative Ramp Rates for Remote Closed- Loop Dispatch and Inverter Autonomous Distributed Control for Colt-VAR in Response to Grid-Simulator Step Changes in Voltage ............................................................................... 39 Table A-1. Scenarios Names and Descriptions ........................................................................................... 46 10 1 Introduction Distribution voltage control is fundamental to delivering power system stability and ensuring the high-level of power quality required by end-user equipment. Historically, the challenge for distribution utilities has been ensuring voltage drop from the distribution substation—i.e., where transmission voltages (e.g., 69 kV and greater) are stepped down to distribution primary voltages (e.g., typically 4–35 kV)—to the customer does not bring voltages outside character-of-service ranges. Utility character-of-service voltages are typically tied to ANSI C84.1 Range A, which stipulates utility distribution voltages be within +/-5% of nominal (American National Standards Institute, n.d.). Appliances have a wider band and designed to operate within +/- 10% as specified by the National Electrical Manufacturers Association, allowing for voltage drop from the meter to an appliance point of common coupling (e.g., a socket). As power flows along utility assets (e.g., transformers and conductors), voltage drops due to the impedance of these electrical assets and their associated thermal losses. Typically, utilities try to keep per unit voltages (e.g., 1 p.u. is 100% of nominal) higher at the distribution primary side of substation transformers (i.e., keeping these upstream voltages closer to 1.05 p.u., the upper range of ANSI C84.1 Range A). Doing so helps account for voltage drop as power flows down to the end user and the meter, ensuring end-user supply stays within bounds. In the United States, a substation typically has on-load tap-changing capabilities, which allows its transformers to increase or decrease voltage relative to nominal (a range of 16 tap positions is not untypical, with each tap accounting for approximately a 0.625% change in voltage, enabling a range of +/- 10% of nominal). Utility voltage control assets downstream of the substation provide additional voltage control, such as (1) capacitor banks, which can use (volts-amps reactive) VARs to increase voltages or (2) line regulators (autotransformers with the same input/output nominal voltage, which can provide tap-changing capabilities). These assets offer utilities enhanced control capability and coordination opportunities, allowing operating voltages on the lower side of established bounds to achieve energy efficiency. Enhanced voltage control through coordination of voltage control assets to achieve power quality or energy efficiency objectives is referred to as volt-VAR optimization (VVO). Utilities can have multiple concurrent objectives with respect to VVO, that include energy efficiency, power quality, and, of interest to this study, solar integration. Utilities are exploring the integration of additional assets to traditional voltage control, which include the use of advanced metering infrastructure (AMI) to provide grid-edge voltage measurements, and smart inverters, which can provide active and reactive power voltage regulation (Figure 1). 11 Figure 1. Traditional and future voltage monitoring and control assets considered in this report. VVO can provide energy efficiency benefits by reducing energy consumption through reducing voltages within utility-specified character-of-service bounds. This strategy is generally referred to as conservation voltage reduction (CVR) (Kennedy & Fletcher, 1991; Wang & Wang, 2014). As voltage decreases, end-use loads consume less energy. There are three theoretical ideal responses, described below, for load responses to changes in voltages, which overall can be characterized by an exponent form equation for voltage response, see (1). (1) Though there are no ideal loads in reality, most loads exhibit responses that are close to those of ideal load models (McKenna & Keane, 2016b). For an ideal load model, voltage response can generally include the following types of response: • Constant Impedance (P=V2/R): Reducing voltage reduces power to the square (i.e., in Equation (1, np is equal to 2). Resistive heating and older style incandescent and halogen lighting closely followed these forms of response; most traditional CVR benefits came heavily from incandescent lighting before these were phased out for compact fluorescent and light-emitting diode lighting solutions. • Constant Current (P=VI): Reducing voltage reduces power directly proportional (i.e., in Equation (1, np is equal to 1). Motor loads most closely follow constant current. 12 • Constant Power (P=VI): Reducing voltage increases current (i.e., in Equation (1, np is equal to 0). Modern power electronic loads more closely follow constant power type responses. For loads that are close to this response, as voltage decreases, current increases, which can increase other distribution I2R losses. Aggregate voltage response depends on the myriad loads connected and consuming power from the distribution system at any given time (McKenna & Keane, 2017). The power and current response of each ideal load type with respect to voltage is presented in Figure 2. In general, with the phaseout of incandescent and halogen lighting and their replacement with compact fluorescent light and LEDs, and the transition to more power electronic interfaces, the voltage response of the load has been decreasing over time. In general: • For np greater than 1, demand will reduce with a decrease in voltage. • For np less than 1, demand will increase with a decrease in voltage. Figure 2. Constant impedance (Z), constant current (I), and constant power (P) active power and current responses with respect to voltage CVR strategies generally have the following effects on losses and demand: • Losses: As a general principle of power transmission, higher voltages result in lower per unit power delivered losses (P=VI: Increase V reduces I, P=I2R: Current proportional to the square of the voltage). CVR typically decreases overall losses but increases percentage losses (i.e., per unit power delivered, losses increase). • For voltage reduction achieving a reduction in demand: o CVR increases the losses per unit power delivered, and, o CVR decreases the overall magnitude of losses. 13 • However, if load does not reduce with a reduction in voltage (i.e., constant power loads): o CVR increases the losses per unit power delivered, and, o CVR increases the overall magnitude of losses. This work focuses on how solar integration and solar hosting capacity1 impact and are impacted by VVO. Overall, integration of solar photovoltaics (PV) causes two initial challenges: (1) the potential of grid-edge overvoltages due to active power injections and (2) the potential for reverse power flow resulting in overloading (in the case of very high PV integration) and protection concerns (Seguin et al., 2016). In this work, we see that VVO can increase hosting capacity by decreasing voltages which increases the headroom before overvoltages are experienced. Overall, the potential to achieve, or improve, energy efficiency and hosting capacity depends critically on the initial network conditions. Networks with existing low voltages will have low opportunity to further reduce voltages; indeed, voltages might have to be increased, reversing the VVO benefits in favor of improving power quality and bringing voltages into character-of- service ranges. Networks with high voltages have higher potential for achieving energy reduction and increasing hosting capacity. The number of control assets (e.g., regulators and capacitor banks) also improves the potential opportunity for VVO. Figure 3 conceptually presents the benefits VVO may bring for networks with initial low or high (with respect to nominal and character-of-service bounds) voltage conditions. Networks with high voltages can achieve benefits of reducing energy consumption, reducing peak demand, and increasing solar integration potential by lowering voltages within character-of-service ranges (McKenna & Keane, 2016a). Networks with low voltages that introduce VVO may increase energy consumption, peak demand, and lower solar integration potential, but they can reduce the risk of voltage excursions. Figure 3. Information graphic on how original network conditions can determine the type of benefits VVO can bring 1 Solar hosting capacity is installed capacity that can be safely integrated onto a distribution network without upgrades. 14 Of interest in this research was how higher levels of solar PV might impact traditional VVO benefits; this is dealt with in detail in later sections. An additional key interest of this research was the potential benefits of integrating smart inverter (SI) dispatch (i.e., active and reactive power dispatch) into a centralized VVO scheme. SI voltage control has been established in recent standards, such as IEEE 1547-2018, and it is being required in certain states in public utility commission rulings (California Public Utilities Commission (CPUC), 2017; Hawaiian Public Utilities Commission (PUC), 2016; IEEE Standards Coordinating Committee 21 on Fuel Cell, Photovoltaics, Dispersed Generation, and Energy Storage, n.d.). Previous work has demonstrated the benefits of SI voltage control and benefits in terms of grid integration (Giraldez et al., 2018; Giraldez Miner et al., 2017). The overall benefits of SI-VVO integration include: • Grid-Edge Visibility: SI technology can give VVO visibility on grid-edge voltages similar to AMI. • VAR Control: SI can provide VAR control in the form of power factor, volt-VAR, and volt-watt with distributed control and the potential for centralized control with communication. AMI integration was also examined in this work. AMI gives VVO grid-edge voltage visibility, which can be thought of as providing greater quantification of grid conditions and can fine-tune VVO control (McKenna et al., 2022). Traditionally, VVO will conservatively reduce voltages in order to allow for headroom from measurements on which VVO relies and which are typically on the primary network. For example, if VVO has a line voltage measurement on the primary network remote from the substation, it may regulate that voltage a few percentage points above the lower bound of character of service to account for voltage drop to end-of-line on the distribution secondary (e.g., regulating to 0.97 p.u. to allow for a 3% drop of the lower ANSI C84.1 Range A limit of 0.95 p.u.). AMI may complicate VVO control and may result in VVO taking the following actions in the following cases: • No Headroom or Grid-Edge Violations Seen in AMI Sata: VVO will raise voltages, thereby reducing energy reduction gains. • Additional Headroom to Reduce Grid-Edge Voltages Seen in AMI: VVO will lower voltages, thereby increasing energy reductions. Also, VVO simultaneously seeing AMI grid-edge measurements above and below character-of- service bounds could further complicate control. Overall, AMI integration should be seen as a positive means of providing accuracy for the best control actions VVO should take. This project examined Utilidata’s VVO algorithms performance both in isolation and in conjunction with other technologies such as solar integration, smart inverters and AMI, and it examined the effects on four National Grid distribution feeders. Three principal tasks were completed as part of this project: 4. Development of a simulation model to assess the power flow benefits of VVO 5. Assessment of how to approach cost-benefit analysis of VVO and SI integration 6. Laboratory demonstration and analysis of centralized versus distribution control for SI volt-VAR control. 15 Simulation models and cost-benefit analysis are presented in Section 0, and laboratory demonstration of centralized versus distribution control for SI integration are presented in Section 3. 2 Modeling and Cost-Benefit Analysis of VVO and SI Dispatch This section details the modeling and simulation performed for the cost-benefit analysis of VVO with SI dispatch. NREL configured and ran multiple scenarios on an OpenDSS power flow model of four suburban distribution feeders that captured both the primary (13.2 kV) and secondary networks (120/240 V). The OpenDSS model was developed using utility GIS and CYME International power flow models, and it was validated against feeder SCADA (supervisory control and data acquisition) and AMI data sets (active and reactive power, and voltage). Synthetic distribution secondaries were added to the model, validating power flow at the grid-edge by using AMI voltage measurements. In these simulations, VVO provided enhanced voltage control with the objective of lowering voltages to maximize the available peak demand and energy consumption reduction. The objective of VVO also includes minimizing voltage excursions outside of ANSI C84.1, which stipulates voltages remain within +/-5%. The objective of minimizing voltages also serves to aid solar integration because the typical first constraint on accommodating solar PV are system overvoltages that result for PV active power injection. Utilidata’s traditional VVO system controls substation load tap changers, voltage regulators, and capacitor banks. The VVO uses a machine learning algorithm to establish relationships between control devices, and it optimizes voltage asset dispatch. This analysis introduces PV inverter grid support functions, such as volt-VAR and volt-Watt, in conjunction with VVO operation, and it includes scenarios where VVO dispatches PV inverters and uses AMI measurements to give end-of-line visibility. The overall setup of the VVO and control assets is presented in Figure 4. The rest of this section is organized into key questions pertaining to VVO cost-benefit framework and simulation analysis, with a focus on PV inverter grid support function integration. 16 Figure 4. Schematic of a feeder using VVO control and the voltage assets being controlled VVO has the potential to both reduce energy consumption and increase hosting capacity by increasing the headroom before an overvoltage threshold is reached. The opportunity to achieve, or improve, these metrics depends on the initial network conditions. Networks with existing low voltages will have low opportunity to reduce voltages; indeed, voltages may have to be increased, reversing the VVO benefits in favor of improving power quality and bringing voltages into character-of-service ranges. Networks with high voltages have higher potential for achieving energy reduction and increasing hosting capacity. The number of control assets (e.g., regulators and capacitor banks) also improves the potential opportunity for VVO. Figure 5 conceptually presents the conditions where VVO will have a greater opportunity to reduce voltages and energy consumption and potentially increase hosting capacity. Networks with existing low voltage conditions may also benefit by bringing voltages back into bounds but still ensuring maximum energy reduction and hosting capacity benefits. Multiple scenarios were run to provide insights into the key research questions in the rest of this section; these scenarios are detailed in in Appendix A. 17 Pre-Existing Low Pre-Existing High Customer Customer Voltages Pre-VVO Voltages Pre-VVO Few Voltage Low opportunity to Medium opportunity Control Assets reduce voltages/ to reduce voltages/ (e.g. just LTC) energy consumption energy consumption VVO Opportunity and increase hosting and increase hosting capacity capacity Low opportunity to High opportunity to Many Voltage reduce voltages/ reduce voltages/ Control Assets energy consumption energy consumption (e.g. regulators, and increase hosting and increase hosting capacity capacity capacitor banks) Figure 5. Opportunity for VVO depending on original network voltages and the number of voltage control assets 2.1 What should be considered in a benefit-cost analysis framework for VVO? The benefit-cost analysis (BCA) assesses the cost and benefits of VVO to integrate and manage distribution voltage control assets and distributed energy resources (DERs) on the distribution network. The objectives of the BCA are to: 1. Propose an initial framework on what should be considered in a full-fledged BCA and how to include stakeholder (i.e., customer, utility, solar developer) costs and benefits. 2. Quantify benefits and costs of selected metrics (Figure 7, page 20). 3. Include a brief discussion of the metrics that are not quantified in this study and how they might be examined with expanded analysis (Section 2.7, page 31). 4. Include a brief discussion of caveats of the scope of this study, such as considerations when contemplating scaling or generalizing the results (Section 2.7, page 31). The BCA considers multiple scenarios to estimate the incremental costs and benefits of using VVO by itself or in combination with AMI, smart inverters using custom volt-VAR curves, volt- VAR activation, and VVO direct inverter dispatch. The BCA uses results from NREL’s simulations and information about electricity cost for different stakeholders to evaluate utility, customer, and solar asset owner costs and benefits, including substation energy imports (consumption), substation energy exports, network I2R losses, and PV curtailment. NREL engaged diverse stakeholders, including utilities, developers, and other nonprofit member organizations and advocates in developing the BCA framework. 18 The overall simulations included SCADA, AMI, DER forecasts, DER technical information, and standard-as-is voltage control asset settings (for comparison to VVO); these were in addition to network GIS and Synergi feeder models that were used to build the OpenDSS power flow model. For price inputs, retail prices, wholesale locational marginal prices and cost of network upgrades (e.g. installing new voltage regulators) were used (Figure 6). Figure 6. Diagram of inputs to the BCA calculation The BCA was performed from two main perspectives, shown in Figure 7. The first is the perspective of residential and commercial customers, customers with solar, and PV system owners (i.e., utility-scale plant owners under a power purchase agreement; see left side of Figure 7). The second is the utility perspective, for which benefits are ultimately passed back to customers. Each perspective has its own set of associated metrics, represented by gray and blue rectangles in the figure. The blue rectangles represent metrics that depend on the results from the power flow simulations NREL has conducted. Each metric can represent a benefit or a cost. For example, for solar customers, curtailment will always represent a cost when it occurs, whereas changes in energy consumption may be a cost (when energy consumption increases) or a benefit (when it decreases). In the upper part of Figure 7 are metrics that can be quantified through our analysis with information from either from simulation results or from the utility. The metrics in the lower region of the figure require additional analysis that may require extensive laboratory analysis (e.g., calculating impact of increased inverter throughput on lifetimes), utility and regulatory inputs (e.g., reducing distribution-use-of-system [DuoS] charges or PV compensation), or analysis out of scope of this project (e.g., hosting capacity analysis). Each metric is described in Table 1. 19 Figure 7. BCA framework for VVO with key metrics from the perspectives of customers, solar owners, and utilities Table 1. Description of VVO BCA Metrics and Their Expected Cost and Impact Metric Description Anticipated VVO Impact/Cost Change in Energy VVO changes the voltage profile on the network, and High consumption correspondingly, customer energy consumption will change due to load voltage dependency. VVO aims to bring voltages down to the lower end of ANSI C84.1 Range A (95%–105% of nominal), which would help reduce energy consumption and thus help reduce customer bills. PV Curtailment PV curtailment can occur due to VAR-priority for volt-VAR Low curves and through activation of volt-Watt functions. Keeping voltages within ANSI C84.1 VVO can help reduce PV curtailment. Inverter Inverter communications are needed for VVO dispatch of Uncertain Communications inverters, but the cost of communications may be incurred through other requirements for communicating with inverters. Inverter communications may provide stacked value (e.g., DER generation visibility, end-of-line voltages, DER program verification, and forecasting), making it unlikely the cost of communications should solely be costed in a VVO BCA. Interconnection DERs may cause distribution networks to hit their limits, Medium Upgrade Costs resulting in the need for network upgrades. VVO may alleviate or defer upgrades by reducing the risk of voltage violations (e.g., overvoltages caused by PV integration in particular), which may reduce interconnection costs. Inverter Lifetime Inverter lifetimes can be impacted by increased VAR Low Reduction throughput used in advanced inverter functions such as 20 Metric Description Anticipated VVO Impact/Cost volt-VAR and volt-Watt. In studies cited in this report, increased VAR throughput for volt-VAR functions increased total inverter kVAh throughput by less than 1%, meaning this impact is likely negligible. In direct VVO reactive power dispatch scenarios, if VVO aggressively dispatches inverters, this [increase] may become more significant. Conversely if VVO keeps voltages within Range A, the increase may decrease inverter throughput by requiring fewer VARs from inverters. Reduced DuoS DuoS charges help utilities recover the fixed costs from Uncertain Charges (future) fixed assets and operation of the network. If VVO can reduce or defer the need for upgrades, DuoS charges may decrease for customers. PV Compensation If PV is used for providing network ancillary services (e.g. Uncertain (Solar Value Stack) is dispatched by the VVO), the benefit from those services could be paid back to customers. Future examination of how to calculate the benefit from PV ancillary services and how to create tariff structures/DER programs for reimbursement is required. Traditional Upgrades Similar to DuoS charges, if VVO can reduce or defer the Medium need for network upgrades, these costs for customers may reduce. VVO Implementation VVO requires communication systems, servers/cloud Uncertain Cost computing, and vendor software support. The costs for these are referred to as implementation costs. AMI/VVO Integration AMI integration to VVO involves communication of meters Uncertain Cost to a distribution management system or centralized system that hosts the VVO algorithm. Additional computing power will also be required to process the large volume of meter data. Integration of AMI data to a distribution management system will provide the utility stacked value in terms of grid-edge visibility and voltage and power quality management; this requirement may impact how these costs are levelized or included in the VVO BCA. Device Lifetime Impact VVO dispatches voltage control assets such as load tap Low changers, regulators, and capacitor banks. Increased switching of load tap changers and regulators may increase wear and tear on these assets and may reduce their lifetimes. Similarly, increased switching and use of capacitor banks may also reduce their lifetimes. VVO algorithms may include constraints so as not to increase the frequency of use of voltage control assets. If VVO significantly changes the operation of voltage control assets, the potential impact on device lifetime, and the accelerated need and cost for replacements, should be included in the VVO BCA. The impact on regulators and load tap changers are more significant than capacitor 21 Metric Description Anticipated VVO Impact/Cost banks because transformers have higher costs than capacitor banks. Energy Supply/Imports VVO impacts distribution feeder imports from the bulk High from Substation (Load system through changing energy consumption and losses + Losses – PV) by changing feeder voltages and potentially impacting PV curtailment. If there is headroom for voltage reduction, VVO should reduce energy consumption, losses, and PV curtailment, which should in turn reduce overall bulk system imports. Reduction in VVO impacts on reducing energy consumption and Low Emissions/Pollution substation imports will also impact emissions associated with energy generation. By reducing voltages and energy consumption, VVO can reduce energy generated by fossil- fired power plants. Increased Hosting VVO aims to bring voltages down to the lower end of ANSI Medium Capacity C84.1 Range A. PV integration hosting capacity is typically limited due to system overvoltages and reverse power flow (and more specifically overloading from reverse power flow). Overvoltages are usually experienced before overloading. By reducing voltages to the lower end of Range A VVO can create additional headroom and accommodate more of solar before overvoltage occur and resulting in increased solar hosting capacity. Enhanced Voltage VVO provides enhanced voltage regulation and can High Regulation reduce the frequency of voltage violations and excursions outside character of service, or ANSI C84.1 Range A. This provides customers benefits in terms of improved power quality and the utility benefits in terms of reducing penalties from poor character of service. 22 By lowering system voltages, VVO can reduce customers’ energy consumption, which would translate into savings. In NREL’s simulations, VVO reduced customer consumption by 1%–2%. VVO can also reduce energy losses in the system. Reductions in customer consumption and system losses lower energy demand from the wholesale market, potentially reducing locational marginal prices, although we expect such reduction to be small. VVO does provide equitable savings for all in terms of reduced energy consumption. However, benefits in terms of reducing asset deferral and increasing hosting capacity are expected to benefit more both affluent customers and customers who consume more energy, because of the nature of retail tariffs (i.e., how fixed costs are spread across customers). 2.2 What is the value of VVO of the examined feeders? VVO provides benefits in terms of reduced energy consumption and reduced losses in the distribution network. VVO, in the simulations for this report—in addition to reducing residential consumption 1%–2%, as mentioned—reduced energy losses seen by the substation by up to 3.3%. Table 2 shows the decrease in substation energy imports when comparing 2019 scenarios with and without VVO. The combined effects of the reduction in losses and customer consumption translates into annual substation import reductions of up to 1,100 MWh and cost reductions of up to 1.6%. These numbers are positive results for this feeder, as before VVO network voltages were at or just below unity, meaning the original potential for voltage reduction was reasonably small. The large number of voltage control assets on this feeder also meant there was originally a relatively tight voltage band, and VVO could take advantage of the number of voltage control assets (e.g., regulators and capacitor banks) by reducing voltage drop and keeping voltages in a tight band in the lower range of ANSI C84.1 Range A. Table 2. Capacity of Rooftop and Non-Rooftop PV for Different Scenarios Year PV Capacity Rooftop (MW) and PV Capacity Non- Total PV Capacity Percentage of Customers with Rooftop (MW) (MW) Solar (%) Base (2019) — — 8.660 2035 + 2.24 MW + 7.874 18.747 (5.2% of customers) Electric Power + 6.3 MW + 12.874 27.800 Research (14.6% of customers) Institute (EPRI) Hosting Capacity Scenario VVO may alleviate the need for upgrades, particularly network upgrades that would be required for voltage power quality. For all scenarios, both in low and higher solar cases, VVO achieves a reduction in energy consumption and can closely control the voltages to the lower end of Range A. In both 2019 and 2035, PV reduces the cost imports at a higher rate than megawatt-hour reductions due to day/night locational marginal price differences (Table 3). 23 Table 3. VVO Resulting in a 1%–2% Reduction in Cost of Energy Imports in 2019 The difference in the 2019 and 2035 reductions in energy imports is principally due to increased solar generation. For 2035, VVO achieves a reduction in energy imports of approximately 1%–2%. For the base case scenario, Figure 8 shows the components of the -1.2% change in substation imports in 2019. Of the 1.2% change, VVO reduces energy consumption by approximately 1,100 MWh annually. Even though annual losses marginally increased by 101 MWh, customer load decreased by 1,109 MWh. Increase Decrease Total 200 - Change in Energy (200) (MWh) (400) (600) (800) (1,000) (1,200) Losses PV Production Customer Load Figure 8. Losses, PV production, and customer load components of the change in substation imports in the Base Case with and without VVO in 2019 24 2.3 How might increased deployment of PV systems change the value of VVO? The reduction in bulk power import costs at the substation was 0.1%–0.2% higher in 2035 than in 2019. The greatest source of substation power import reductions was PV production from new PV capacity added between 2019 and 2035. Most new PV capacity added came from independent producers (88% of total new capacity), not utility customers. This new PV capacity in the system reduced substation import costs by 16%. Table 4 and Table 5 provide the substation net exports and cost savings for VVO in terms of substation imports. Table 4. Volume of Customer Net Exports in 2035 in Annual Megawatt-Hours 2035 Upgraded 2035 Upgraded 2035 Original 2035 Original Network Network with VVO Network Network with VVO Residential customer energy exports (MWh) 1,464 1,473 1,464 1,471 C&I customer energy exports (MWh) - - - - Utility-scale PV exports (MWh) 15,726 15,726 15,726 15,726 Table 5. VVO Cost Savings for 2035 Red squares show the comparison of equal scenarios with and without VVO between 2019 and 2035. Energy reduction achieved just by VVO in 2035 is highlighted in blue squares. 2.4 How will advanced inverter functions from smart inverters impact the value of VVO? Standard SI settings (such as IEEE Category B volt-VAR, customized volt-VAR, or proposed National Grid volt-VAR curves) have the aim to reduce the overvoltages and undervoltages. The total energy consumption may either increase or decrease depending on whether VARs from SI settings are required to mitigate overvoltages (in which case you would see a decrease in energy consumption) or undervoltages (in which case you would see an increase in energy consumption). VVO was observed to reduce energy consumption by 1.887% in the 2035 scenario for unity power factor settings. This number decreased to 1.761% for customized volt- VAR settings where VARs were used to mitigate undervoltage, and energy consumption reduced 25 by 1.889% for IEEE Category B and proposed National Grid volt-VAR inverter setting where most of the VARs were used to reduce high voltages relative to customized volt-VAR setting. The following key observations were made for advanced inverter functions: • VVO controls the voltage to largely within the deadband of SI volt-VAR curves, thus reducing the need for inverter VARs and reducing inverter throughput (Figure 9, page 28). • For the studied feeders, volt-VAR curves provided little additional benefit due to the tight voltage regulator from VVO and the number of voltage regulators and capacitor banks on the network. • Volt-VAR curves did not counter the objective of VVO, but they can provide support at the edge of ANSI C84.1 Range A for networks with particularly high or low voltages. That was not the case for the feeders studied here. 2.5 How can AMI enhance the value of VVO? Advanced metering infrastructure (AMI) data provide greater visibility of how networks perform in terms of asset loadings and voltages, which should help VVO make better decisions. However, integrating AMI also comes with challenges. • Including AMI data from all customers at each time step increases complexity for VVO and communications may contribute to latency. Additional computing power (e.g. servers or cloud resources) would be required to process the AMI data. • AMI infrastructure may not be present for most customer premises for the utility. • Real-time AMI data management can pose greater challenge for utility to keep it reliable (in terms of keeping the system running 24-7 and making sure collected measurements are not erroneous, this will require real-time processing for quality control and assurance) and scalable. • The cost of integrating AMI in VVO in real-time is likely to be high, meaning stacked value applications may be required recoup the cost. Targeting a subset of meters (e.g. those at end-of-line and electrically closest and furthest from the substation) may reduce the cost of AMI integration and still provide high value for the VVO. 26 2.6 Can VVO increase the amount of solar that distribution systems can accommodate? How can advanced inverter functions impact solar penetration? Figure 9 compares the distribution of the percentage of customer voltage hours in different voltage (per unit values or p.u.) ranges for different scenarios. VVO demonstrates that it can bring the voltage down significantly (comparing orange colored bar with bar of other colors) compared to scenarios with traditional voltage control. Lower voltages create greater headroom before overvoltages occur and limit PV hosting capacity. The impact of smart inverters on shifting customer voltages is relatively smaller compared to impact of VVO in general. VVO controls the voltage to a tight band reducing the need for inverter VAR provision. With VVO, voltages are within the deadband (more than 90% of the time) or in a region where percentage VAR contribution (for aggressive volt-VAR) from PV system is very small (5% to -10%) and from other curves (e.g. IEEE Category B and proposed National Grid volt-VAR curves) is zero. In general, SI settings help reduce voltage above 1.0 p.u. compared to unity power factor simulation and will help boost PV hosting capacity of the feeder. VVO gives additional voltage headroom before the upper limit of 1.05 p.u. of ANSI C84.1 is reached. This is shown in Figure 10 (page 29), which shows the headroom from customer voltages and the upper voltage limit. For 2035, the simulations did not show any significant voltage violations with both VVO and no VVO scenarios being able to accommodate the additional PV capacity. An additional scenario was examined using the numbers from the Electric Power Research Institute (EPRI) using their hosting capacity numbers produced from the EPRI Drive tool. This added an additional solar capacity of just over 9 MW on the feeders compared to the 2035 case. In this scenario overvoltages were experienced on the primary network without VVO, but adding VVO control was able to mitigate those overvoltages, see Figure 11 (page 30). In summary, the studies found: • VVO reduces voltage to the lower end of ANSI I84.1 Range A which allows for additional headroom until overvoltage limits are reached. Solar PV integration typically results in overvoltages first and thermal overloading second. VVO may defer the need for upgrades by allowing additional PV capacity before overvoltage limits are reached. • For the feeder examined, overvoltages are not expected with or without VVO up to their 2035 DER generation forecasts. However, VVO creates additional voltage headroom for accommodation solar capacity. • Running a scenario using National Grid provided hosting capacity values for the circuit, overvoltages were experienced on the primary without VVO. VVO was able to bring these back into bounds enabling the hosting capacity limit to be reached. • For the feeders analyzed advanced inverter functions, such as volt-VAR and volt-Watt were not required as VVO tightly controlled the system voltages to within the deadband of these functions. For other circuits, with higher voltages and less voltage control assets, advanced inverter functions might provide more support than for the feeders analyzed here. 27 No VVO – Volt-VAR curves: higher Note VVO operates voltages in the dead-band reducing the need for VAR support VVO – lower voltages Figure 9. Percentage of customer voltage hours in different voltage bands among various scenarios VVO reduces customer voltages achieving energy reduction, reducing the need for inverter VAR support, and increasing the headroom until overvoltages (which helps increase system hosting capacity). 28 Difference to overvoltage limit VVO increases the No VVO – headroom before higher overvoltages are voltages VVO – experienced, helping lower increase hosting voltages capacity. Figure 10. Percentage of customer voltage hours for the difference between actual voltages and 1.05 p.u., the high voltage limit, among various scenarios VVO increases voltage headroom for accommodating solar integration. 29 No VVO results in over-voltages at non-rooftop site connected to primary network VVO reduces in over-voltages at non-rooftop site connected to primary network Figure 11. Voltages at 5-MW non-rooftop site connected to primary network in the hosting capacity case study, examining voltages with and without VVO VVO reduces overvoltages, allowing the increased solar capacity. 30 2.7 Considerations for Implementing a BCA for Utilities This study was conducted for National Grid feeders, and such the following considerations are for those specific feeders. However, some of the general lessons for implementing the BCA could be applied to many other utilities. The results from this study are specific to the feeders examined, and different feeders will yield different potential saving opportunities and benefits from VVO. To expand on the BCA done as part of this project in a way that reflects the scope outlined in the National Grid BCA Handbook version 3, future analysts could take the following points into consideration: • The analysis period considered in the present study was one year. To comply more precisely with the Handbook, analysts would need to expand the analysis period to a longer span and include the time-value of money. The period should be agreed upon with the utility and other stakeholders. • Such analysis would also need to consider future effects on rates. Additionally, it would also need to consider effects on entire service territory with a diverse set of feeders and capture impacts on bulk power system (e.g. wholesale market prices, transmission losses, etc.). Some feeders may have greater benefits than the feeder analyzed here and some may have lower, in general, Figure 5 (page 18) presents the theoretical opportunity depending on the feeder voltage profile and number of voltage control assets. • Such analysis would also need to forecast future wholesale prices (i.e. for substation imports). The price of solar PV and how that would impact the rate of adoption, and potential changing price of integrating VVO and communications. • Figure 7 (page 24) shows metrics that were not quantified in this study. Examples include inverter lifetime reduction due to an increase in throughput, changes in interconnection study costs (performed by the utility and paid by developers), considerations about whether and how solar customers and independent PV producers could be compensated by the benefits that they may provide to the grid, and how VVO could help increase hosting capacity or delay the need for grid upgrades, among others. Each of these topics could be a research project on its own and analysts looking at these costs and benefits may want to spend resources quantifying them. • The impact of VVO on customer energy equity and justice could also be considered. The expectation is that all customers stand to benefit if VVO can reduce voltages and achieve energy consumption reductions for all. Hosting capacity benefits are likely to only benefit those who can afford to install solar PV, although ancillary benefits from additional renewable energy deployment, such as improved air quality, should benefit all. Reduced or deferred asset upgrades from VVO may provide uneven benefits due to how fixed costs are recouped in existing retail tariff structures. 31 3 Laboratory Integration of SI Communication Laboratory testing set out to demonstrate remote dispatch of SI communication. Testing goals were to establish and compare centralized and distributed SI control for voltage regulation. Centralized distribution control of smart inverters would demonstrate how these assets could be integrated into a utility voltage control regulation (e.g., such as VVO control scheme), coordinating with other traditional control assets such as tap-changing transformers, capacity banks, line voltage regulations, etc. The overall purpose was to help understand the challenges of a centralized VVO dispatching SI active and reactive power. The research in this section was also the subject of a conference paper (Mahmud, 2023). The testing objectives were to: • Demonstrate the successful operation of the proposed centralized SI communication pathways. • Compare centralized remote dispatch (open and closed-loop) versus distributed inverter- based dispatch (distributed autonomous) to understand response times and characteristics. • Demonstrate SI dispatch operating under real-time power hardware-in-the-loop (PHIL) testing. The testing aimed to demonstrate and validate a communication setup from a remote SCADA dispatch (Distribution Test Manager [DTM]) using the OrionLX and SEL-RTAC to the two 250 kW solar inverters. The inverters were demonstrated to be able to receive active and reactive power dispatch commands. 3.1 Overview of Testing for Remote Open-Loop, Remote Closed-Loop Control, and Distributed Autonomous Control NREL used Triangle MicroWorks’ Distributed Test Manager (Triangle DTM) as the centralized controller to dispatch the inverter (two 250-kW solar inverters). The Triangle DTM was programmed with volt-VAR curves to remotely sense inverter terminal voltage and issue corresponding reactive power dispatch. This was contrasted with inverter-programmed volt-VAR to compare centralized and distributed volt-VAR testing response times and characteristics. Open-loop and closed-loop testing were demonstrated, these entailed: • Open-Loop: In open-loop testing, the Triangle DTM was used to issue reactive power (Q-dispatch) commands to the inverter. • Closed-Loop: In closed-loop testing, the Triangle DTM was programmed to receive inverter terminal voltage and correspondingly issue reactive power (Q-dispatch) commands to the inverter. • Distributed Autonomous: The overall test setup for open-loop and closed-loop centralized testing is given in Figure 12. The Triangle DTM is used to dispatch the SI. The OrionLX is used as to represent a utility substation gateway, and the SEL-RTAC was configured as an outstation device. 32 SI Communications Path Power Hardware Testing DNP3 Smart DNP3 (slave) (master) Inverters Triangle SEL PV DNP3 Orion LX DNP3 RS-485 DTM RTAC Simulator Modbus Programmed Server Volt-VAR AC Response Ethernet Test kit (VVO AC Junction emulator) Box SI and Grid Simulator AC Xformer Grid Simulator Legend: DC power AC power Modbus DNP3 Figure 12. Setup for open-loop and closed-loop centralized SI dispatch response time characterization The additional goals of this testing were to understand: • SI controls: How to set diverse volt-VAR curves in the inverter? How to send commands and receive confirmation from the inverter? How to communicate PQ 2 dispatches (or power factor) to the inverter? What is the response time/ limitations/ characterize performance? • Integrate SI to VVO: How to integrate communications to a VVO? Establish communication protocols. What limitations should the VVO be aware of? Establish communications successfully. • Test and characterize integration: Test, in PHIL setup, inverter integration, characterize range of communications abilities and confirm successful VVO integration. Table 6 presents the key tests and comment on the results. Based on the testing goals just outlined the tests were successful in all cases. One caveat was when setting custom volt-VAR curves, off-center volt-VAR curves were not possible (i.e., the volt-VAR curves set points had to be symmetrical). Table 6. Key Tests and Findings Based on Testing Goals Test Result Note Program the inverter for custom Volt- ü Custom volt-VAR curves can be configured, but VAR curves off-center volt-VAR curves are not possible Ability to receive inverter terminal ü Inverter essential power measurements were voltage able to be transmitted (e.g., voltage, active power, reactive power) Ability to issues active power dispatch ü Active power commands were able to be command dispatched Ability to issue reactive power dispatch ü Reactive power commands were able to be command (for Watt and VAR-priority) dispatched 2 Define PQ in a footnote. 33 Remote closed-loop and open-loop dispatch involves a number of steps from inverter detection, communication to the centralized control, and issue of dispatch. Characterization of the response time was separated into several elements: • Detection Time (Td): time needed for the equipment under test (EUT), in this case the inverter, to measure the parameter (e.g., inverter terminal voltage) • Uplink Communication Latency (TCU): time needed to transmit the measured parameter (e.g., inverter terminal voltage) from EUT terminal to the external controller (i.e., the Triangle DTM) • External Controller Computation Time (TCC): time needed for the external controller to compute the control signal (i.e., inverter VAR dispatch) for the EUT based on the measured parameter • Downlink Communication Latency (TCU): time needed to transmit the external controller generated control signal measured from external controller to EUT terminal • EUT Response Delay (TR): time needed by the EUT to respond to the control signal • Ramping time (Tramp): time needed by the EUT to change the response from one step to another step • Communication and Computation Loop Time (TL): total needed to measure the EUT parameter, communicate the parameter to the external controller, compute external control signal and transmit the control signal to EUT • Tracking Error: difference between the control signal and actual EUT response. Figure 13 shows the communication pathway elements between the external (centralized/hierarchical) controller and the EUT. Error! Reference source not found. illustrates the communication elements in time-series and the tracking error. 34 Figure 13. Overview of communication elements involved in centralized dispatch Figure 14. Overview of time-series elements in communication elements in remote dispatch and tracking error The total communication lag time is the sum of each individual element, as presented in equation (2). 𝑇𝐿 = 𝑇𝑑 + 𝑇𝐶𝑈 + 𝑇𝐶𝐶 + 𝑇𝐶𝐷 + 𝑇𝑅 (2) 35 3.2 Test Results Open-loop and closed-loop dispatch were used to calculate communication element times and characterize the inverter response. 3.2.1 Remote Open-Loop Dispatch The open-loop dispatch outline procedure and test result were as follows: • An open-loop test, issuing a reactive power dispatch command from the DTM to the inverter and determining the reactive power ramp rate in response to this external control signal. • In this test, the positive reactive power ramp rate was found to be 4.16%/s while the negative reactive power ramp rate was found to be -14.24%/s. The results are displayed in Figure 15. • The ramping was observed to be linear, with the inverter manufacturer confirming a linear ramp was expected from an external dispatch. (a) (b) Figure 15. Open-loop tests and corresponding inverter ramp rates for (a) a positive ramp rate and (b) a negative ramp rate 36 3.2.2 Remote Closed-Loop Dispatch The open-loop dispatch outline procedure and test result were as follows: • A closed-loop tests involved the inverter and the Triangle DTM. The grid simulator was used to issue a step change in voltage. This terminal voltage was communicated from the inverter to the Triangle DTM, a corresponding reactive power dispatch was calculated and issued to the inverter. This test was used to calculate the total communication and computation loop time (TL). In these tests, the DTM was programmed with a volt-VAR function. • In this test, for a step change of voltage from high to low, the T L was found to be 11.243 seconds and for step change of voltage from low to high, the TL was found to be 16.1 seconds. The results are displayed in Figure 16. • It should be pointed out here that the Measured TL reported here were based on a limited number of tests. To get a comprehensive characterization of T L, more experiments would be needed. (a) (b) Figure 16. Closed-loop tests and corresponding inverter ramp rates for (a) a positive ramp rate and (b) a negative ramp rate 3.2.3 Remote Closed-Loop Compared to Distributed Autonomous Inverter Control The next set of tests were used to compare closed-loop testing, dispatching the inverter according to a volt-VAR curve, and distributed autonomous testing using inverter-programmed volt-VAR curves. The grid simulator was used to provide step changes in voltage to see how remote closed-loop dispatch using the Triangle DTM and distributed autonomous inverter program volt- VAR dispatch responded. Figure 17 shows the results for step changes in grid-simulator voltage for inverter autonomous distributed volt-VAR control, showing both the voltage and inverter reactive power. 37 Figure 17. Distributed autonomous volt-VAR control (i.e., inverter-programmed volt-VAR control) responses to step changes in grid-simulator voltage Figure 18 shows results for remote closed-loop volt-VAR dispatch of the inverter, showing both the voltage and inverter reactive power, again with voltage step changes being controlled by the grid simulator. Figure 18. Remote closed-loop volt-VAR inverter dispatch (i.e., Triangle DTM inverter dispatch) responses to step changes in grid-simulator voltage The key observations in comparing these two controls are provided in Table 7. The key findings are as follows: • Overall, remote closed-loop dispatch had a linear and faster ramp rate than autonomous distributed inverter control. Inverter autonomous control follows a first-order control response, which was expected by the manufacturer. This raises a question on how inverters should response to external control signals, both in the speed and characteristic of the response. • Autonomous distributed inverter control started responding to voltage step changes in under a second, whereas due to communication delays, remote closed-loop testing took on the order of 11.7 seconds with a standard deviation of 4.78 seconds. The remote closed-loop communication delay was not constant throughout testing. 38 • The mean ramping time (Tramp) for remote closed-loop dispatch the ramp rate was observed to be in the range of 6.3%/s – 16.26%/s and -4.5 %/s to -14.05 %/s. For autonomous distributed (inverter based) control, the ramp rates included inverter embedded control response and were slower than centralized control and ranged between 0.45 %/s and -0.49 %/s. Table 7. Estimated Communication Lag Time and Positive and Negative Ramp Rates for Remote Closed-Loop Dispatch and Inverter Autonomous Distributed Control for Colt-VAR in Response to Grid-Simulator Step Changes in Voltage Response Times and Metric Centralized Control Inverter Controla Ramp Rates (units) TL (second) Mean 11.74 0.77 Standard deviation 4.78 0.31 Maximum 20.16 1.20 Minimum 2.80 0.20 b Positive ramp rate Mean 10.87 0.45 (%/second) Standard deviation 3.68 0.13 Maximum 16.26 0.58 Minimum 6.37 0.15 c Negative ramp rate Mean -10.02 -0.49 (%/second) Standard deviation 2.96 0.01 Maximum -4.50 -0.47 Minimum -14.05 -0.51 a For Inverter Control, 𝑇𝐶𝑈 = 0; 𝑇𝐶𝐶 = 0; and 𝑇𝐶𝐷 = 0 b Rate of change of reactive power from low value to high value c Rate of change of reactive power from high value to low value 3.3 Overview of Power Hardware-in-the-Loop Testing Power hardware-in-the-loop (PHIL) tests were used to further characterize remote closed-loop and distributed autonomous inverter volt-VAR control. The overall test setup is provided in Error! Reference source not found., and key test elements are described below: • Real-Time Simulation: A distribution network model was used to run in real-time. The same feeder models used in Section 2 were converted to ePHASORsim and run in an OPAL-RT to simulate the feeder in real-time. • Load and PV Profiles: Separate profiles for all the loads and PV generation for the simulated PVs were used in the real-time simulation distribution feeder model. • Legacy Voltage Regulating Device Control: Time-series control set points (e.g., on/off status of capacitor banks, line regulator tap positions) for legacy devices were used from simulations run (Section 2). 39 • PV Simulator Control: A PV simulator was programmed with irradiance profile for the duration of tests and took an analog signal from the Opal-RT. • Grid-Simulator Control: Inverter terminal voltage obtained was obtained from the point of common coupling node from the real-time simulation of the network and used as driver signal for the grid simulator. • Transformer: A 600 V/480 V transformer was used to step down the inverter terminal voltage to connect to the grid simulator. The overall tests were able to demonstrate that: • Voltage Regulation: Voltage regulation could be achieved by both a centralized control inverter dispatch (remote closed-loop) and distributed inverter (autonomous inverter) control. • Impact of Delay: The impact of the delay to respond to a changing voltage (e.g., TL) as well as the reactive power ramping rate has impact on the network voltage regulation. External Controller Triangle Programmed DTM Volt-VAR Response Test kit (VVO DNP3 emulator) Real-Time Simulation Opal-RT Orion OPAL-RT DNP3 LX SI Communication Server OPAL-RT Analog I/O DNP3 ePhasorSim (Feeder Model) DNP3 SEL RTAC Using time-series control set points for Analog legacy devices RS-485 Signal SI and Grid Simulator Smart DC PV Inverters Simulator Analog Signal AC Junction Box AC Xformer Grid Simulator Legend: DC power AC power Modbus Analog Signal DNP3 Figure 19. Setup for PHIL tests 40 However, these tests mainly focused on; the interoperability of the external controller and the inverter, and demonstration of an experimental setup (via the scenario analysis) to examine the performance of centralized and distributed voltage regulation. Detailed performance analysis of any centralized or distributed controller using the setup will need additional experiments. Overall, the testing demonstrated the communication paths for a centralized VVO system to dispatch inverters using SCADA communications via the OrionLX and SEL-RTAC, see Figure 20. The team was able to configure and validate the communications. The delays in communication and response time may be of concern for voltage control and there may be advantages to using a hybrid communication architecture (i.e., remote control and dispatch, but the ability to receive inverter measurements update remote settings from a centralized controller). How inverters respond to remote dispatch signals (e.g., the type of ramp; linear, first- order, and the timing) is an area for further research and may need to be designed for-purpose. Inverters could easily be programmed to respond with a first-order or linear response or respond faster or slower than other utility regulation equipment. (a) Inverter active and reactive power and terminal three-phase voltages for unity power factor (baseline) 41 (b) Inverter active and reactive power and terminal three-phase voltages for distributed autonomous inverter volt-VAR control (c) Inverter active and reactive power and terminal three-phase voltages for remote closed-loop volt-VAR inverter dispatch Figure 20. Inverter active and reactive power and three-phase voltages for (a) unity power factor, (b) distributed autonomous inverter volt-VAR control, and (c) remote closed-loop dispatch control for a PHIL test demonstrating the setup depicted in Error! Reference source not found. 42 4 Conclusion Three principal tasks were completed as part of this project, which were to: (1) develop a simulation model to assess the power flow benefits of VVO, (2) assess how to approach cost- benefit analysis of VVO and SI integration, and (3) laboratory demonstration and analysis of centralized versus distribution control for SI volt-VAR control. Overall, the key findings of this project were: • VVO can provide energy efficiency benefits by reducing voltages within character-of- service ranges and reducing energy consumption and losses; benefits identified in this report that are specific to the feeders analyzed ranged between 1% and 2%. The VVO opportunity is determined by existing voltage ranges. o VVO also has the potential to decrease the need for network and traditional upgrades by improving the voltage profile of the circuit, avoiding network under and over-voltages. • VVO has additional benefits under high solar adoption scenarios. VVO can help mitigate solar integration voltage control challenges, such as high voltage ramps due to net load changes and mitigation of overvoltages. Energy efficiency benefits can increase under VVO, as VVO can mitigate voltage rises from solar that would increase energy consumption. • VVO mitigating potential over-voltages from solar PV has the potential to expand the capability of a circuit to accommodate more solar capacity, dependent on existing voltage conditions. • The benefits of SI integration into VVO will heavily depend on the degree to which utility voltage control assets have already been deployed. For the scenarios studied in this report, a high number of voltage control assets (e.g., load tap changers, capacitor banks, and line voltage regulators), which enhanced VVO benefits, left little room for additional SI voltage control benefits. For feeders with less utility voltage control assets, smart inverters are likely to have greater benefits than seen in this report. • Integration of smart inverters into a centralized voltage control scheme was demonstrated through external remote control and communication for dispatching a smart inverter in a laboratory setup at NREL’s Energy System Integration Facility. This was contrasted to a distributed autonomous control by the inverter, communication lags from central dispatch and different ramp rates and response characteristics (linear versus first-order for centralized and autonomous respectively) were observed. o Autonomous distributed inverter control were shown to respond to voltage step changes in under a second. Remote external dispatch, due to communication delays, took on the order of 11.7 seconds with a standard deviation of 4.78 seconds. o The delays in the communication and response times might be of concern for voltage control, and there might be advantages to using a hybrid communication architecture (i.e., remote control and dispatch but with the ability to receive 43 inverter measurements to update the remote settings from a centralized controller). For record of the CRADA final report this work did not result in any subject inventions or record of inventions (ROIs). Subject Inventions Listing: None ROIs: None 5 References American National Standards Institute. (n.d.). C84. 1-1995. Electric Power Systems and Equipment Voltage Ratings. California Public Utilities Commission (CPUC). (2017). Electric Rule 21. Giraldez, J., Hoke, A., Gotseff, P., Wunder, N., Blonsky, M., Emmanuel, M., Latif, A., Ifuku, E., Asano, M., Aukai, T., & Sasaki, R. (2018). Advanced Inverter Voltage Controls: Simulation and Field Pilot Findings. NREL Publications, October. www.nrel.gov/publications. Giraldez Miner, J. I., Nagarajan, A., Gotseff, P., Krishnan, V. K., Hoke, A. F., Ueda, R., Shindo, J., Asano, M., & Ifuku, E. (2017). Simulation of Hawaiian Electric Companies Feeder Operations with Advanced Inverters and Analysis of Annual Photovoltaic Energy Curtailment. Hawaiian Public Utilities Commission (PUC). (2016). Rule 14H. https://www.hawaiianelectric.com/Documents/my_account/rates/hawaiian_electric_rules/14.pdf IEEE Standards Coordinating Committee 21 on Fuel Cell, Photovoltaics, Dispersed Generation, and Energy Storage. (n.d.). {IEEE} 1547-2018 -{IEEE} Standard for Interconnection and Interoperability of Distributed Energy Resources with Associated Electric Power Systems Interfaces. Kennedy, B. W., & Fletcher, R. H. (1991). Conservation Voltage Reduction (CVR) at Snohomish County PUD. IEEE Transactions on Power Systems, 6(3), 986–998. Mahmud, Rasel, Subhankar Ganguly, Jing Wang, Killian McKenna, and Ning Li. "Experimental Analysis of Distribution Network Voltage Regulation Using Smart Inverters." In 2023 IEEE 50th Photovoltaic Specialists Conference (PVSC), pp. 1-7. IEEE, 2023. McKenna, K., Gotseff, P., Chee, M., & Ifuku, E. (2022). Advanced Metering Infrastructure for Distribution Planning and Operation: Closing the loop on grid-edge visibility. IEEE Electrification Magazine, 10(4), 58–65. 44 McKenna, K., & Keane, A. (2016a). Distribution network voltage control for peak load alleviation. 2016 IEEE International Conference on Power System Technology, POWERCON 2016. https://doi.org/10.1109/POWERCON.2016.7754033 McKenna, K., & Keane, A. (2016b). Electrical and Thermal Characteristics of Household Appliances: Voltage Dependency, Harmonics and Thermal RC Parameters. http://hdl.handle.net/10197/7538 McKenna, K., & Keane, A. (2017). Open and closed-loop residential load models for assessment of conservation voltage reduction. IEEE Transactions on Power Systems, 32(4), 2995–3005. Seguin, R., Woyak, J., Costyk, D., Hambrick, J., & Mather, B. (2016). High-Penetration PV Integration Handbook for Distribution Engineers. In NREL - National Renewable Energy Laboratory. https://doi.org/10.2172/1235905 Wang, Z., & Wang, J. (2014). Review on Implementation and Assessment of Conservation Voltage Reduction. IEEE Transactions on Power Systems, 29(3), 1306–1315. 45 Appendix A. Description of Scenarios Table A-1. Scenarios Names and Descriptions Scenario Name Description 2019 Upgraded Network No VVO control using feeder models with additional network upgrades (more regulators, capacitor banks and load rebalancing). This is for 2019 load and solar levels, see Table 2. 2019 Original Network No VVO control using the original feeder models (less regulators, capacitor banks than upgraded network). This is for 2019 load and solar levels, see Table 2. 2019 Upgraded Network VVO control using feeder models with additional network upgrades with VVO (more regulators, capacitor banks and load rebalancing). This is for 2019 load and solar levels, see Table 2. 2019 Original Network with VVO control using the original feeder models (less regulators, VVO capacitor banks than upgraded network). This is for 2019 load and solar levels, see Table 2. 2035 Upgraded Network No VVO control using feeder models with additional network upgrades (more regulators, capacitor banks and load rebalancing). This is for 2035 load and solar levels, see Table 2. 2035 Original Network No VVO control using original feeder models (less regulators, capacitor banks than upgraded network). This is for 2035 load and solar levels, see Table 2. 2035 Upgraded Network VVO control using upgraded feeder models with additional with VVO network upgrades (more regulators, capacitor banks and load rebalancing). This is for 2035 load and solar levels, see Table 2. 2035 Original Network with VVO control using original feeder models (less regulators, VVO capacitor banks than upgraded network). This is for 2035 load and solar levels, see Table 2. 2035 Upgraded Network VVO control using feeder models with additional network upgrades with VVO and CatB (more regulators, capacitor banks and load rebalancing). This is for 2035 load and solar levels, see Table 2. Smart inverters are set using IEEE Category B volt-VAR curves. 2035 Upgraded Network VVO control using feeder models with additional network upgrades with VVO and PQ Dispatch (more regulators, capacitor banks and load rebalancing). This is for 2035 load and solar levels, see Table 2. Inverters reactive power are dispatched by VVO. Hosting Capacity Limit No VVO control using feeder models with additional network Upgraded Network upgrades (more regulators, capacitor banks and load rebalancing). This is for solar capacity deployed equal to that in published National Grid capacity maps for these feeders, see Table 2. Hosting Capacity Limit VVO control using feeder models with additional network upgrades Upgraded Network with VVO (more regulators, capacitor banks and load rebalancing). This is for solar capacity deployed equal to that in published National Grid hosting capacity maps for these feeders, see Table 2. Hosting Capacity Limit VVO control using feeder models with additional network upgrades Upgraded Network with VVO (more regulators, capacitor banks and load rebalancing). This is and Agg. VVar for solar capacity deployed equal to that in published National Grid hosting capacity maps for these feeders, see Table 2 (page 23). Smart inverters are set with an aggressive volt-VAR curve without a deadband, see Figure 9 (page 28). 46