5

Models for Distribution System Planning

Susan Tierney, Analysis Group, moderated a session on models for distribution system planning. The speakers were Jason Fuller, Pacific Northwest National Laboratory (PNNL); Roger Dugan, Electric Power Research Institute (EPRI); Aleksi Paaso, Commonwealth Edison (ComEd); Colton Ching, Hawaiian Electric; and John Lee, Xcel Energy.

Tierney opened the session by pointing out that distribution planning is on the cutting edge, where many of the biggest challenges and opportunities facing the electric power system lie, such as bidirectional power flow, innovative devices, and new customer behaviors. “The practitioners in the field are really hitting the ground running in terms of addressing a lot of new activities,” she said. “The question is whether the models are keeping up or what’s needed to evolve more effectively to make sure that the tools are … up to the tasks for distribution planning.”

Speakers discussed existing modeling capabilities and emerging needs as distribution systems transition to incorporate greater dynamism and respond to evolving technologies, policies, and consumer behavior.

JASON FULLER, PACIFIC NORTHWEST NATIONAL LABORATORY

Fuller, integration team lead for the Electricity Infrastructure Group at PNNL, discussed existing tools and future opportunities for distribution planning, including as it relates to other connected systems. “It’s not just the grid anymore,” he said. “As power engineers, we have a tendency to think about transmission lines and power lines and transformers. … But

our system is becoming something that’s much more intertwined with the fabric of society. It’s becoming tied into communications and other elements of the system.”

To understand power load behavior, PNNL created GridLAB-D, which is used to model load-grid interaction for multiple settings, transitions, and market events. The model also enables utilities to explore how different loads will interact both across the system and for individual users—for example, if energy prices change. In addition, GridLAB-D includes simulations to extrapolate progression and storage needs for multiple time scales, an improvement over static snapshots of system use.

PNNL is also actively modeling microgrids and their dynamic behavior at the individual or networked level. These models show that microgrid use can enhance overall grid stability and demonstrate how combining multiple microgrids could improve system resilience, Fuller said. They also allow insights into low-inertia systems, which can affect system balance, and electromechanical dynamic models for individual generators, entire systems, and their interactions. These electromechanical dynamic models are particularly useful for isolated power systems, where individual behavior can cause unwanted consequences, Fuller noted.

A broader challenge, in Fuller’s view, is to understand the grid’s interplay with other systems, such as communications or building control systems. To address this need, he described how co-simulation approaches that model multiple domains using existing tools, models, and data sets can help users examine different levels of detail, define different domain relationships, and create aggregate results (Figure 5.1). While co-simulation with existing tools is a useful approach now, he also noted the need to think creatively about other approaches going forward. “This is what’s

useful right now … but this is not necessarily the only way moving forward,” he said. “Are there other ways that we might want to look at the system, say, 10 or 15 years from now, that are more comprehensive and more singular in nature in terms of modeling and understanding?”

Fuller suggested several areas for improvement. He noted that an upcoming report, Electric Distribution System Planning with DERs—Tools and Methods, discusses today’s state-of-the-art modeling capabilities and presents ideas for more precise forecasts. In particular, he pointed to the need for electromechanical and electromagnetic models that fill distribution-side gaps for inverter-based technologies, as well as ways to improve behind-the-meter representations of loads and their dynamic system-wide impacts.

He also highlighted the need to transition from hosting capacity tools to integrated, holistic planning tools: “Optimization is good; engineering-based decisions are good. Now, how do you bring those together so they could be in a cooperative system, and make those decisions more holistically?” he asked. Closing, he stressed the importance of enhancing multidomain modeling that incorporates distribution planning with communications, markets, and transportation sectors, as well as replacing deterministic solutions with probabilistic or risk-based modeling.

ROGER DUGAN, ELECTRIC POWER RESEARCH INSTITUTE

Dugan, senior technical executive at EPRI, co-developed OpenDSS, EPRI’s main distribution system analysis research tool. The greatest strength of OpenDSS, in Dugan’s view, is its flexibility: it can incorporate a vast range of factors, such as unbalanced multiphase power flow, quasi-static time series for long-term dynamics, smart inverters, and energy storage models. Built-in parallelization allows parallel processing to be exploited on laptops to speed up simulations. Unlike most distribution system analysis tools, it is also able to produce limited transmission-distribution simulation and integrate dynamics analysis.

Traditionally, Dugan noted, distribution systems analysis was based on simple, efficient snapshot simulations reflecting particular points in time. While useful in many circumstances, snapshots are no longer sufficient to meet the needs of planners and operators in today’s complex environment, he argued. Those traditional models also excluded innovations such as batteries, solar generation, and cloud computing. To incorporate these technologies, EPRI used OpenDSS to develop a new multiyear planning tool, the Adaptive Distribution Assessment and Planning Tool (ADAPT). By providing a dynamic perspective over years or decades, the tool is able to help operators find and fix problems such as overloads or voltage violations and search for solutions, including possible storage solutions.

EPRI has met many of the distribution system analysis goals that the organization envisioned a decade ago, including sequential time simulation, meshed network simulation capability, advanced load and generation modeling, high phase order modeling, integrated harmonics, and user-defined behavior. Remaining goals for future development include better modeling of smart grid controllers and better dynamics simulation capabilities for distributed energy resources (DERs) evaluations. Closing, Dugan noted that having more mature dynamics analysis will greatly improve transmission-distribution co-simulation.

ALEKSI PAASO, COMMONWEALTH EDISON

Paaso is the director of distribution planning, smart grid, and innovation at Commonwealth Edison (ComEd), a utility serving 4 million customers across northern Illinois. He introduced several new ComEd initiatives to prepare for future systems.

First, ComEd is building capacity to accommodate the significant increase in the use of DERs. For example, a recent hosting capacity analysis, evaluating DER installation locations, was incorporated into its annual distribution planning activities. In addition, the utility is considering how to best include nonwire alternatives (NWAs). It has also created a methodology to assign a dollar value to DERs, which, he noted, add value to distribution systems and to systems and markets beyond them. This methodology is being converted to a tool to incorporate alternative solutions into distribution planning, and ComEd is currently pilot testing two battery-storage units based on this output.

ComEd also employs metrics to identify how to take advantage of developments such as renewables and DERs to inform optimal microgrid location and design. “It’s not just the integration of renewables, but it’s also about utilizing those renewables or utilizing that DER to the maximum of its capacity for things like forming microgrids,” Paaso explained. The Bronzeville Community Microgrid (BCM), designed to bring clean, sustainable power to approximately 1,000 businesses and residences in one Chicago neighborhood, is one example. To provide the new modeling capabilities required to design the BCM, ComEd integrated hardware, system models, microgrid controllers, and overall operations controls, enabling interoperability testing and demonstrating how cutting-edge modeling and technologies are changing utility operations. Advanced planning models like these feed into the entire utility system and modernize it for reliability, resiliency, and security, Paaso said. Although incorporating new DERs will always be a challenge, he posited that new planning tools and modeling approaches will emerge to meet it.

COLTON CHING, HAWAIIAN ELECTRIC

Ching, senior vice president of planning and technology at Hawaiian Electric, discussed the unique challenges of providing electric power to islands that operate in isolation from both the mainland United States and each other. Hawaii has embraced renewable energy sources, with one in three homes using rooftop solar and a statewide goal of 100 percent of electricity sales from renewables by 2050. That policy is driving Hawaiian Electric’s plan to determine customer and infrastructure needs; outline large, innovative strategies for grid integration; and implement system-wide modernization procedures, Ching said.

The three distribution planning tools that Hawaiian Electric uses today are (1) LoadSEER, to produce load forecast profiles for different DERs (Figure 5.2), which also includes economic and geospatial variables; (2) Synergi, to evaluate thermal and voltage limits with integrated DER analysis, especially to improve hosting capacity (Figure 5.3); and (3) PSSE and PSCAD, circuit modeling software to model load flows, short-circuit analyses, dynamic stability analyses, and voltage disruptions.

While distribution planning tools were used in the past to identify violations, looking ahead they will increasingly be relied upon to inform solutions for future planning. The changes under way in Hawaii illustrate the need for future distribution planning tools to be expanded and updated to incorporate bidirectional flow, the dynamic nature of loads, new equipment, and new DERs, Ching said. “The theory that the distribution system in the future will have a two-way power flow is being enacted today on our feeders,” he said. “And that change, and all of the

technology and customer choices that are causing this change on our distribution system is really driving how we need to change, how we plan our distribution system going forward.” He added that these developments also create the need for more granular modeling results, not just at the snapshot level but across hours, days, or even years.

In addition, Ching posited that distribution planning itself should be integrated into steady-state and dynamic distribution models, transmission and protection analyses, and generation-portfolio planning. Echoing the suggestions of other speakers, he stressed that risk and probabilistic analyses should replace deterministic analysis and noted the importance of creating tools that allow nontechnical audiences to understand model outputs and how they influence decision making.

JOHN LEE, XCEL ENERGY

Lee is senior director of electric distribution engineering at Xcel Energy, which serves a diverse area from Michigan to New Mexico. Xcel uses multiple tools to optimize distribution planning, such as DRIVE, Synergi, OpenDSS, and SCADA. These enhanced tools help Xcel understand interconnection requirements, bidirectional power flow, and hosting capacity, creating precise and multisourced forecasts.

Lee outlined several challenges to future distribution planning. First, it is important to incorporate modeling into the right place into the workflows of planning engineers, who, understandably, prefer checklists or other simplified tools over complicated modeling programs. “If we can do our integration with a checklist, or a screening tool, or Excel spreadsheets,

they’d rather do that than run a Synergi program to actually do the model,” he said. “The trick is knowing when to make the transition. … When is your penetration level high enough that you really have to go down and do that detailed modeling?” He stressed that new tools must be easy to understand and use if they are to be adopted.

Second, meeting Xcel’s ambitious goal of 100 percent carbon-free resources by 2050 will require intensive integration and modeling of each system in order to maintain a careful balance of rapid and accurate analysis. In addition, new projects, such as community solar gardens, are a large undertaking, requiring sufficient data collection before a model can be run to determine their cost-benefit ratio. NWA analysis, for example, can only tell Xcel if it is cheaper to build a new transformer than use NWAs once it is more widely adopted.

In the future, Xcel plans to implement an advanced distribution management system across the entire service area, which will require large-scale information technology integrations and sufficient feeder data; integrating volt/volt-ampere reactive (VAR) optimization, which will also require feeder data; establishing an advanced metering infrastructure in order to collect that data; creating new field area networks; and improving fault location, isolation, and system restoration.

Pilot projects, in partnership with NREL and EPRI, are under way to analyze grid impact and planning. Xcel is also studying community resiliency in multiple cities to understand the role of microgrids. These projects may not be cost-effective, but they demonstrate future possibilities and provide the necessary data to improve overall planning, Lee said.

DISCUSSION

After the speakers’ remarks, Tierney moderated a discussion that focused on how new distribution planning and analysis tools are adopted in the industry; the interplay of policy, planning, and uncertainty; and training for the next generation of workers.

Adopting New Distribution Tools

Tierney asked if distribution companies and utilities were ready to move beyond deterministic modeling to use different kinds of advanced modeling tools. Lee explained that Xcel Energy tries to make effective use of simple tools, but recognizes that it is first necessary to understand the criteria, make the right assumptions, use the appropriate data, and interpret the results correctly.

Ching added that in his experience, the bigger challenge is for longtime distribution planners to embrace the fact that planning, with today’s

distributed systems and demand response, requires new tools. “If you can get folks to really internalize it, understand what distribution planning is in today’s environment, then I think the use of the tools and adoption of them will come rapidly,” he said. Once they make that leap, they adapt to new tools as quickly as planners who are at an earlier stage of their careers. Fuller agreed that new tool adoption is faster among younger engineers, who as students had access to open source software like OpenDSS, MATLAB, and GridLAB-D, and are generally more open to different perspectives when they enter the workforce.

Paaso noted that ComEd also hopes to increase new tool adoption and is creating a DER planning team to develop new distribution planning tools and processes for its engineers. Dugan reported that EPRI has also begun to encourage the use of dynamics analysis for distribution in its research.

Policy, Planning, and Uncertainty

Morgan asked about the interplay between planning, policy, and device or system interconnection. Fuller answered that one challenge to device integration is that vendors sometimes keep data or models private to protect their intellectual property. Ching added that Hawaiian Electric has trouble meeting DER demand for different customers, and has struggled to find a policy that pleases both commissioners and consumer advocates. Lee said Xcel has similar challenges, and noted that one solution is to use cluster studies, instead of sequential studies, to better model and plan for capacity.

Asked if planning was robust to policy uncertainty, Lee replied that Xcel’s integrated distribution planning includes varying policy scenario analyses. Those results helped inform the decision to move forward with load-zero implementation, although he noted that it was a struggle to incorporate so many different potential scenarios. As a utility, Xcel is generally uncomfortable with too much speculation and prefers to refrain from policy modeling, he said.

Ching added that Hawaiian Electric also struggles with balancing policy and regulation with utility work. He said it has helped to use different metrics for different stakeholders, while still emphasizing the distribution planning impact. Fuller posited that a reversible model, capable of starting with a goal or with a policy, would be advantageous.

Anuradha Annaswamy, Massachusetts Institute of Technology, asked Ching what Hawaiian Electric needs to reach its 2050 goal. Ching replied that due to each island’s unique geography, distributed systems, on top of increased renewable rates, will surely be required. For the time being, planners are fixing known limitations to their protection systems and schemes,

and generally are working to solve for the next short-term period, with the belief that each small step will bring them closer to their goal.

Ching added that the most surprising aspect of Hawaii’s high penetration of renewables was the competitive, high-value creation that has resulted from pairing solar with storage on a large scale. Tierney asked if Hawaiian Electric uses advanced metering infrastructure (AMI), and Ching replied that meters are mostly used as sensors for potential DER systems, not for billing. Lee noted that Xcel is fully deployed with AMI, and Paaso added that ComEd is also moving in that direction.

Training the Next Generation

Elizabeth Wilson, Dartmouth College, asked speakers to comment on needs for training the next generation of workers. “All my students want to do is build renewables,” she said. “What are some of the skills that we at universities should be training them for? How should we be setting our programs up to really cross-train to address some of these emerging issues that are both technical, but also sociotechnical?”

Paaso answered that ComEd has created large-scale skills identification and training programs for its workers, which include a broader understanding of new technology, dynamic studies, 8,760 analyses, and integrating scripting language for daily distribution planning. Ching suggested that universities should update and reposition power utility programs in order to attract engineering students, who might otherwise not consider utility work. Hawaiian Electric, he noted, uses internships and student programs as a way to recruit talented students.

Fuller stressed that engineers must learn to view the world as an integrated system, where narrow problems impact larger plans. “One of the things I’d encourage is getting people to elevate their vision [to] a little bit bigger picture of the world and understanding how this is an integrated system,” he said. “Do not assume that what has worked in the past will always work in the future.”