Distribution System Analysis and the Future Smart Grid

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 6, NOVEMBER/DECEMBER 2011 2343 Distribution System Analysis and the Future Smart Grid Robert F. Arritt, Member, IEEE, and Roger C. Dugan, Fellow, IEEE Abstract—The “smart grid” refers to various efforts to modernize the power grid through the application of alternate sources of energy and intelligent devices. The present national interest in smart grid applications has generated many questions concerning the role of distribution engineering in the future. What features do utility engineers need in distribution system analysis tools to support the future smart grid? This paper will discuss some relevant Electric Power Research Institute research in this area that focuses on selected issues related to smart grid analysis relevant to rural utilities. The essential characteristics of distribution system analysis tools to support analysis of these issues are discussed. Index Terms—Power distribution system analysis, smart grid. I. I NTRODUCTION HE SMART grid means different things to different people. To some, it is an emphasis on communications and control, which have not typically been represented in distribution system analysis. To others, the smart grid means distributed resources—generation, storage, and demand response. These issues have been addressed by many authors since the early 1990s, and many distribution system analysis tool suppliers have already implemented some capabilities to model distributed resources. However, there remains much work to do. The Distribution System Analysis Subcommittee (DSAS) of the IEEE Power and Energy Society (PES) Power Systems Analysis, Computing, and Economics Committee presented a paper at the 2010 IEEE PES General Meeting on this subject [1]. This paper discusses selected subjects from the DSAS paper in the context of distribution engineering in rural electric utilities. The Distribution Test Feeders Working Group (WG) of the DSAS has already done work on related subjects, for example, to address the concern for having large induction generators (e.g., wind turbine generators) on distribution feeders [2]. Part of the motivation for that work was the concern expressed by rural electric utility engineers. The WG is continuing to work in this area and in other areas related to smart grid and distribution system analysis tool development. Distributed generation (DG) is not the only concern. Other perspectives on the smart grid T Manuscript received December 26, 2010; revised August 10, 2011; accepted August 17, 2011. Date of publication September 22, 2011; date of current version November 18, 2011. Paper 2010-REPC-553.R1, presented at the 2011 Rural Electric Power Conference, Chattanooga, TN, April 10–13, and approved for publication in the IEEE T RANSACTIONS ON I NDUSTRY A PPLICATIONS by the Rural Electric Power Committee of the IEEE Industry Applications Society. The authors are with the Electric Power Research Institute, Knoxville, TN 37932 USA (e-mail: [email protected]; [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TIA.2011.2168932 contain an emphasis on such things as extensive monitoring, intelligent protection, microgrids, and energy efficiency. What kind of distribution system analysis framework is needed to support the features being proposed for the smart grid? Will there be a need for distribution system analysis if everything is monitored thoroughly? What can be done if more is known about the system? What different approaches to designing distribution system analysis tools will be required to support this? These are questions that this paper explores. It certainly seems likely that there will be some kind of convergence of distribution system planning, distribution system monitoring, and distribution state estimation (DSE) into distribution management systems (DMSs). Exactly how that happens remains an open question. The Electric Power Research Institute (EPRI) is actively involved with smart grid demonstration projects as well as exploring advanced distribution system analysis concepts. Selected relevant issues are discussed in this paper. II. S MART G RID C HARACTERISTICS Research is just beginning on many issues related to the smart grid, and its features continue to be defined. Those features likely to have an impact on the direction of distribution system analysis include [1] the following: 1) distributed resources: a) generation; b) renewable generation (variable resources); c) energy storage; d) demand response; 2) communications and control: a) advanced metering infrastructure (AMI) deployed throughout the system; b) high-speed communications to metering and controls; c) state estimation; 3) improved reliability: a) automated fault location; b) automated restoration; c) planning (switch locations); d) improved asset utilizations; 4) improved energy efficiency: a) end-use efficiency; b) delivery efficiency; c) at the planning stage; d) operationally (active voltage regulation, etc.). Smart grid issues will accelerate a natural evolution toward more optimization, real-time operation, and intelligent algorithms in distribution system analysis. There is also a need to 0093-9994/$26.00 © 2011 IEEE 2344 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 6, NOVEMBER/DECEMBER 2011 cosimulate power and communications networks for integrated design of power, control, sensor, and communication systems. In the past, data uncertainties have been cited to justify approximate analysis methods. In the future, integrated systems and real-time state estimation will require the best available models. III. S TATE OF THE A RT From the late 1960s to the mid-1980s, distribution system analysis computer programs evolved from simple balancedload voltage drop calculators that automated hand calculations to sophisticated systems with databases and interactive graphics. The system models were still largely based on the simplified radial circuit techniques of the voltage drop calculations. Many power engineers who have been involved largely with transmission system problems and are not intimately familiar with the industry are under the mistaken impression that this is still the state of the art in distribution system analysis. There have been many significant advances since that time. Key vendors began to adopt three-phase models in the 1980s. The rural electric power industry played an important role in the development of this capability with educators and consultants promoting methods such as those described in W. H. Kersting’s book [3]. The basic need that this satisfied was to enable unbalanced feeder modeling. In the 1990s, another driver emerged: DG. DG modeling forced most distribution system analysis tools to have full three-phase modeling capability for both urban systems and rural systems. For those analysis packages that provide it, harmonics analysis was another influential driver. The norm today for distribution system analysis packages is to provide for full three-phase circuit power flow and short-circuit analysis. The following list was published in [1] in an attempt to concisely capture essential features of the present state of the art in distribution system analysis tools. 1) Most distribution system analysis tools can perform full three-phase analysis; some, such as EPRI’s OpenDSS [4], can go well beyond three phases. 2) Most utility distribution system analysis is performed using tools originally designed to assess power delivery at one point in time. 3) A few tools have the capability to perform simulations over periods of time such as a day, week, month, or year. 4) Tools and techniques are designed for uniprocessors, which is generally satisfactory for present needs. 5) Many tools, particularly those designed for the North American market, exploit the typical radial nature of medium-voltage (MV) and low-voltage (LV) distribution systems for certain simulation efficiencies even if they also offer meshed network analysis. 6) Harmonics analysis, if available, is an optional feature. 7) Time-domain packages exist, but frequency-domain packages are preferred for distribution analysis. 8) Dynamics analysis is uncommon with distribution system analysis tools. 9) Distribution planning and distribution operation tools are largely separate modules with differing capabilities. 10) Modeling of the distribution system generally ignores the secondary (LV) distribution system. 11) Modeling of end-use loads is generally with timeinvariant ZIP models. These are necessarily generalizations of the state of the art, and the list is not intended to be comprehensive. Specific packages offer more advanced capabilities in one or more of these areas. The intent is to highlight certain capabilities that could be impacted by needs to analyze smart grid capabilities. What new modeling capabilities will distribution system analysis tools of the future need? The authors have been involved in several smart grid research efforts using the EPRI OpenDSS computer program [4]. This program has a common heritage with distribution system harmonics solution engines and, therefore, has some capabilities that may seem unusual in a distribution system analysis tool. EPRI has made this program available in open source to encourage the evolution of features to provide distribution planners with the tools that they will need to analyze and support the smart grid. Some of the capabilities of this tool that EPRI is encouraging vendors to adopt for supporting its member utility needs include the following. 1) Sequential-time power flow solutions, in various time step sizes ranging from less than 1 s to 1 h, to accommodate analysis of such things as voltage regulation issues stemming from renewable generation and storage. 2) Meshed network solutions are handled as easily as radial circuit solutions. 3) Modeling of controllers is separate from circuit elements, better enabling modeling of various smart grid controller functionalities such as volt–var control of solar photovoltaic (PV) generation. 4) Advanced flexible load and generation modeling. 5) Detailed high-phase order circuit modeling capability to enable analysis of such things as follows: a) neutral-to-earth voltages (NEVs); b) crowded rights-of-way with several circuits; c) atypical, but common, fault conditions such as transmission overbuild falling on distribution. 6) Integrated harmonic solution capability to enable analysis of, for example, NEV (typically fundamental and third) and higher frequencies that might come from inverters. 7) Scriptable behavior to enable modeling of situations not anticipated by the software developers. 8) Dynamics analysis for investigating islanding concerns as well as open-conductor faults involving machines. The relevance of these capabilities should become apparent as these topics are discussed in the following sections. IV. ROLE OF D ISTRIBUTION S YSTEM A NALYSIS What kinds of analyses will distribution engineers want to perform for the future smart grid? In some visions of the smart grid, the distribution system is saturated with voltage, current, and power monitors with all the data being available online in a matter of seconds. Some who have this vision foresee a more limited role for distribution planning than today. They suggest ARRITT AND DUGAN: DISTRIBUTION SYSTEM ANALYSIS AND THE FUTURE SMART GRID that distribution system analysis tools will seldom be needed if engineers can simply sit at their desks and obtain the voltage and power consumption at every customer. Planning becomes more of a straightforward bookkeeping problem. On the other hand, as has happened for many technological advances, there could be a greater need for detailed distribution system analysis to better manage all the technology. Functions such as reconfiguration after an emergency will likely still require significant distribution system analysis capability. With widespread monitoring, the analysis may be simpler if one is able to easily tabulate available paths to find one with sufficient remaining capacity. Of course, this function needs to be fast, since the implementation is real time, and it must account for such things as missing data due to failed communications channels. Modeling analysis is likely to remain a strong component of the reconfiguration function. EPRI’s vision is that distribution planning and DMSs with access to real-time loading and control data will converge into a unified set of analysis tools. That is, real-time analysis and planning analysis will merge into common tools. Distribution system analysis tools will continue to play an important role, although they might appear in a much different form than today. While some may think of rural electric utilities as lagging behind in technological development, the small-to-moderate size of these utilities can make it more practical to implement smart grid technology system-wide than it is for large urban utilities. Therefore, it is expected that some rural electric utilities will be among the early adopters of advanced DMS tools. V. M ODELING C APABILITIES N EEDED FOR S MART G RID The following section highlights a few of the key needs. A. Modeling for DG The addition of DG to the electrical distribution system has been one of the key drivers in the evolution of distribution system analysis tools over the last 15 years. Three-phase circuit modeling and other advanced features have been added to accommodate the needs of DG modeling. Introducing DG into existing systems requires that it be carefully integrated with the power system operating practices. The key concerns include the following: 1) voltage rise and regulation; 2) voltage fluctuations; 3) protective relaying and control functions; 4) impact on short-circuit analysis; 5) impact on fault location and clearing practices; 6) need for an interconnection transformer; 7) transformer configuration; 8) harmonics; 9) response to system imbalances such as open-conductor faults due to failing splices. The authors are actively involved in research projects in highpenetration DG. Of particular interest as of this writing is solar PV generation. One possible result of higher penetrations of DG on the smart grid is that distribution system designs will evolve to better accommodate DG. For example, voltage regulators 2345 Fig. 1. Simulation of regulator response to solar ramp function. may be set a few percent lower to allow for voltage rise (as well as drop) due to DG output. If not, situations like that shown in Fig. 1 will occur for solar ramping events. Solar PV generation is subject to cloud transients. When a cloud obscures the sun, the PV output ramps down, resulting in a drop in the voltage. After timing out, the regulators tap up to correct for the drop in voltage. When the cloud passes, the PV output ramps back up with the regulator tap too high for the power level. The voltage is pushed over 105% when the solar output recovers from a cloud transient until the regulators again compensate by tapping down. This analysis requires a series of solutions at a time step of 1 s and an accurate regulator model. A typical finding in our ongoing research in high-penetration PV is that utility voltage regulator targets will have to be set 1%–2% lower to provide sufficient “headroom” to accommodate this kind of power swing. DG analysis will likely require further enhancements to distribution system analysis tools, such as better meshed network analysis for modeling spot networks, dynamics analysis, etc. This is already apparent by observing the new features showing up in the major distribution system analysis packages. B. Time Series Simulations The ability to perform time series simulations like that shown in Fig. 1 will be a key feature of software to support the smart grid. Few of the present commercial tools were designed to efficiently perform such simulations and manage the large amounts of result data that this analysis produces. Most distribution system analysis tools were originally designed to support the traditional planning analysis of designing to meet peak demand. Thus, the main task was to solve the power flow for one specific point in time: the predicted peak demand. The smart grid, as envisioned by many, will have varying generation, dispatchable generation, controllable loads, and controllable voltage regulation devices. Solving for only peak demand can fail to properly reflect the actual annual load profile. 2346 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 6, NOVEMBER/DECEMBER 2011 The time increment will be different for simulations of various smart grid features. Some of the issues and typical time step sizes are as follows [1]: 1) 2) 3) 4) 5) 6) 7) 8) electric vehicle charging (minutes, hours); solar and wind generation (seconds); dispatchable generation (minutes to hours); storage simulations (minutes to hours); energy efficiency (hours); DSE (seconds, minutes); end-use load models (minutes to hours); end-use thermal models (minutes to hours). The duration of these simulations will be from a few minutes to days and years. In any case, key requirements for distribution system analysis software will be as follows: 1) to perform fast solutions; 2) to capture and process voluminous results. C. Modeling for Imbalances The differences between a symmetrical component model and an unbalanced phase-domain model can yield quite different results. A symmetrical component model uses only the positive- and zero-sequence impedances to represent overhead and cable line segments as balanced impedances. However, asymmetries in the mutual coupling between adjacent phases, adjacent feeders, and conductors yield impedances that are not balanced between phases. Distribution System Analysis (DSA) has come a long way over the years because most distribution system analysis tools can perform full three-phase analysis; however, few programs exist that can go beyond three phases. More and more of the circuits that the authors have analyzed include multiple feeders sharing right-of-ways with as many as 17 conductors on the same pole sharing a common neutral (as well as several communications messengers). As an example, consider two heavily loaded long feeders sharing the same pole construction and static wire like that shown in Fig. 2. This should be modeled with at least six coupled conductors to properly model the coupling between the two adjacent feeders. The currents computed for each phase using a detailed 6 × 6 impedance matrix are shown in the figure. Fig. 3 shows the results computed for a positive-sequence model of each feeder. All line currents in the symmetrical component model come out balanced. Fig. 4 shows the differences between these two models for the phase voltages computed at bus A (at the end of feeder A). In the detailed model, the voltages vary from 0.972 p.u. on phase A to 1.026 p.u. on phase B, which may prove to be too much imbalance for some three-phase loads. In the positivesequence model, the computed phase voltage is approximately 1.0 p.u. on all three phases, which would indicate no problems for three-phase loads. The impacts of unequal phase impedances become a greater issue as line currents are increased on the feeders sharing common construction over a significant distance. This is just one illustration of a coupled set of conductors requiring more than three-phase solutions to get the correct Fig. 2. Heavily loaded feeder A and feeder B sharing the same pole construction with full-phase model results. Fig. 3. Symmetrical component model of feeder A and feeder B. Fig. 4. Bus voltage at the end of radial feeder A. answer. Several situations arise in NEV simulations where there may be over 15 conductors on a pole (see the NEV test case in the IEEE Test Feeders [7]). ARRITT AND DUGAN: DISTRIBUTION SYSTEM ANALYSIS AND THE FUTURE SMART GRID 2347 D. Large Systems It is common for distribution planners to model only one feeder at a time. One thing that is almost certain to occur from the implementation of the smart grid is the need to model several feeders—or even several substations—simultaneously. At a minimum, adequate tools should be able to represent all feeders fed from a common bus. It is frequently not possible to capture the true benefits of distributed resources, or the full extent of operating problems that might occur, without this model. Another issue requiring this capability would be the representation of switching between feeders. Various other proposals require modeling of a defined “distribution planning area” that includes a few substations. To accomplish this, tools would have to be able to accommodate 10 000–100 000 buses. The authors routinely model circuits consisting of 5000– 10 000 buses. There is a project currently scheduled to begin in 2011 that will require a 100 000-bus model, and inquiries have been received about capabilities to solve an 800 000-bus model. While million-bus models may seem far-fetched now, expected advances in computer technology could make this practical in just a few years. Parallel computing could be one approach. Developing algorithms for dividing the problem into a series of smaller ones is another approach. When doing such analyses as simulating distribution automation over several substations, it is an advantage if one is able to model a large part of the system all at once. Fig. 5. Computed and AMI voltages at customer during peak week. single sourced and radial. After connecting DG units to the distribution system, this assumption is no longer valid. Also, DG is often protected with voltage relays and multifunction relays that are monitoring quantities other than overcurrent. Sometimes, the protection devices are communicating to other control devices. This is often best analyzed by event-based simulation of the entire protection and control system response, which is not always easy to represent with TCC plots. This is another function that will require time series simulation. The time step size would be in milliseconds. E. DSE DSE will be a key feature of distribution system analysis software intended to support smart grid applications. DSE could enable real-time optimization, adaptive protection and control, pricing signals for demand response, and many other smart grid features. Widespread deployment of AMI, sensors, and automated devices will provide more data than ever before, so that robust state estimation becomes more feasible. Transmission state estimators are well developed, but those techniques do not all apply well to distribution systems. Some of the barriers to DSE include the following: 1) low X/R ratios; 2) phase imbalances; 3) prevalence of current magnitude, voltage magnitude, and demand interval measurements; 4) communication latency and bandwidth; 5) nonsimultaneous samples; 6) still not enough measurements to make the feeder observable. DSE will be a key component of DMSs. It is also likely to become a key component of distribution planning. In fact, DSE, DMS, and distribution planning functions are expected to merge. F. Protective Relay Coordination Simulation Conventional protection coordination is based on timecurrent curves (TCCs), assuming that distribution systems are G. AMI Load Data The improvement in metering data accompanying the anticipated expansion of AMI and other smart grid applications will provide better inputs to distribution system models. The ability to collect load data over a long period of time is critical to understanding a circuit’s behavior. An improved model will provide better data on end-use patterns and diversity factors for better quantification of distribution system efficiency and improving automation simulations. Efforts are currently underway to use the AMI data for both inputs to and verification of distribution models. Typically, substation metering power data are used to allocate loads and provide the yearly load shape information. As AMI data become available, each load can be allocated separately with its own annual load shape. This provides additional insight to the circuit; however, this is easier said than done as the AMI data are often not synchronized with other AMI metering on the same feeder, sampling rates may vary from minutes to hours, and the vast amount of data slows computation time. EPRI has begun work on incorporating AMI data into DSA. In the EPRI’s Green Circuit initiative, AMI data were available on selected circuits. Fig. 5 shows an example of using AMI data to compute a customer’s voltage during the peak week. The raw AMI data at the customer in this example were sampled at a 15-min interval and are compared to the voltages computed in the model using an hourly load shape derived from AMI data. As shown in Fig. 5, reasonable results can be computed and 2348 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 6, NOVEMBER/DECEMBER 2011 verified through the use of the customer’s AMI data; however, much work still remains in determining the optimal usage of AMI data. H. Modeling Controllers Controllers are a key component of smart grid visions. Whether smart or dumb, they can have significant impact on the solution. Controllers in common usage today include capacitor switch, load tap changer (LTC), voltage regulator, tie switches, reclosers, sectionalizers, and breaker controls. Various visions of the smart grid would add such things as generator dispatch, energy storage control, microgrid control, electric vehicle control, and demand response control. Modeling of controllers is weak and inconsistent in today’s distribution system analysis tools. For example, it is common to assume that the substation LTC can correct the voltage to the desired level without actually simulating it to see if it is possible. Also, the state of controllers is not easy to determine with a static power flow solution. It is frequently necessary to simulate the daily load shape to get all the voltage regulation devices (principally capacitor and regulators) to arrive at the proper values. The desire to have this type of simulator is likely to increase with the introduction of more controlled devices in a smart grid implementation. I. Modeling Communications Communications bandwidth and latency are topics of high interest. What good does it do to have the distribution system saturated with monitors and controllers if it is not possible to communicate with them fast enough to achieve an improvement over existing system designs? This issue is currently being studied at several institutions, e.g., [11] and [12]. Simulating the latency has significant implications for distribution system analysis tools. While loading simulations can be adequately performed with time increments no less than 10–15 min, latency simulations involving multiple controllers and monitors would ideally be performed in time intervals of a few seconds or even millisecond intervals. Simulating large systems for a day, week, month, or year at such a small interval would place severe demands on distribution system analysis tools and the computer systems on which they would run. This is a challenge for any simulation where small time steps are required over long periods of time, e.g., millisecond time steps for months. However, it may be necessary to get the right answer. Advancements in parallel computing will hopefully enable this simulation. J. Work Flow Integration At the same time that life as an engineer is becoming more complicated with smart grid implementation, the time pressure to deliver has also increased. Some utilities are being inundated with DG applications for multi-megawatt projects, and the regulatory requirement to complete an impact study is typically one month. In some cases, screening decisions must be made in two weeks. No matter what new analysis capabilities are de- veloped, the utility engineer cannot spend weeks learning new research-grade software tools and then more weeks developing separate models for each tool. The new tools have to be robust and integrate seamlessly with corporate data systems and other software systems, such as the geographic information system now used at most utilities. VI. N EXT S TEPS A. Advancing Distribution System Analysis Tools Many of the expected advancements in distribution system analysis tools will come directly from the commercial vendors as their user communities demand new features. The IEEE will also play a role. Groups such as the IEEE PES DSAS play a significant role in advancing tools by making sure that the analysis and modeling needs of the smart grid can be met. The DSAS fills this role as follows: 1) by developing new benchmarks that stretch the capabilities in various ways (see the next section); 2) by supporting data exchange standards such as the following: a) IEC 61968 and the Common Information Model [14]; b) National Rural Electric Cooperative’s MultiSpeak [15]; 3) by producing a recommended practice that defines many of the terms, quantities, and procedures used in distribution system analysis, i.e., IEEE Standard P1729, “Recommended Practice for Electric Power Distribution System Analysis”; 4) by organizing paper and panel sessions as well as other means of technology transfer to keep the industry informed of advances and needs. B. Test Feeders The Test Feeders WG of the DSAS has published several test feeders [7] and is in the process of developing new cases. These test feeders will be central in the effort to verify which distribution system analysis software is suitable for the simulation needs for smart grid modeling. The test feeders will also be drivers of advancements in distribution system analysis technology, being designed to incrementally stretch the capabilities of the tools. As an example, the WG has recently introduced an 8500-node test feeder [9] to help benchmark the ability of software and proposed analysis algorithms to handle larger circuits like those encountered in many rural locations (Fig. 6). Distribution planners today commonly work with system models consisting of several thousand buses. Given the needs identified for supporting smart grid applications, planners will want to model even larger systems. Therefore, any algorithm intended for distribution system analysis tools of the future must scale up from the small system models used to test the algorithm to many thousands of elements and buses. The 8500-node test feeder has also been designed to present challenges to distribution system analysis software which are ARRITT AND DUGAN: DISTRIBUTION SYSTEM ANALYSIS AND THE FUTURE SMART GRID 2349 Additional test feeders planned by the WG include the following: 1) NEV; 2) short-circuit benchmarks; 3) DG protection; 4) larger distribution system models; 5) inverter-based DG models; 6) asymmetrical contingency test feeder. VII. C ONCLUSION Fig. 6. One line of the 8500-node test feeder circuit. Fig. 7. Residential load configuration in the unbalanced secondary model. common in smart grid modeling. The challenges include the following: 1) constructing models of large unbalanced distribution feeders; 2) solving large distribution systems containing numerous imbalances; 3) modeling the 120/240-V center-tapped transformer common in North American systems (see Fig. 7); 4) modeling LV (secondary) distribution; 5) heavily loaded systems that are near convergence limits; 6) advanced controls. The 8500-node test feeder includes many elements that may be found on a North American rural distribution feeder: multiple feeder regulators, multiple switched capacitor banks, secondaries, and service transformers. While the likely initial use of the test feeder will be to simply prove that a method can solve the power flow for the defined loads in an acceptable amount of time, the test feeder was also selected for its potential for serving as the basis for future advanced test feeders. Two examples for which there is presently interest are as follows: 1) distribution automations, including voltage and var control simulation; 2) annual load shape simulation for evaluating energy efficiency options, renewable generation, and electric vehicle impacts. The ability to accurately model distribution systems with smart grid components and associated behaviors will require distribution system analysis tools to evolve significantly to meet the challenges. Key challenges include the following: 1) merging planning and real-time analysis; 2) very large system models; 3) system communication simulation; 4) handling a large volume of AMI data; 5) AMI-based decision making; 6) time series simulations; 7) DSE; 8) detail modeling (service transformers and service wiring); 9) distribution models, including the effects of multiple feeders, transmission, and subtransmission systems; 10) DG integration and protection; 11) generator and inverter models; 12) regulatory time pressures; 13) control systems and control system interactions (i.e., DMSs, distributed controllers, distributed resource controls, etc.); 14) modeling of intelligent end-use devices and systems (i.e., smart appliances, demand response, price response, etc.). Newer faster computing methods will continue to be a key emphasis of tool development in distribution system analysis to support the smart grid. The ready availability of multiprocessor computers will likely play a key role. Software to exploit these machines for distribution system analysis needs to be developed. This may require looking at distribution system analysis methods in completely different ways than the way that it is done today. While this paper deals mostly with simulation capabilities, user interface will have to evolve with the capabilities. EPRI is cooperating with and helping lead IEEE efforts in assessing distribution software tools and developing test benchmarks. The creation of new test feeders will include the expanding capabilities of distribution system analysis software tools, particularly those capabilities necessary to support smart grid components and behaviors. R EFERENCES [1] R. C. Dugan, R. F. Arritt, T. E. McDermott, S. M. Brahma, Sr., and K. Schneider, “Distribution system analysis to support the smart grid,” in Proc. IEEE Power Energy Soc. Gen. Meeting Conf., 2010, pp. 1–8. [2] R. C. Dugan and W. H. Kersting, “Induction machine test case for the 34-bus test feeder-description,” in Proc. IEEE PES Gen. Meeting, Jul. 2006, pp. 1–4. 2350 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 6, NOVEMBER/DECEMBER 2011 [3] W. H. Kersting, Distribution System Modeling and Analysis. Boca Raton, FL: CRC Press, 2007. [4] OpenDSS Program, Available on the Internet Through SourceForge.net. [Online]. Available: http://sourceforge.net/projects/electricdss [5] T. E. McDermott, “Working group on recommended practice for distribution system analysis—P1729,” in Proc. IEEE PES Gen. Meeting, Pittsburgh, PA, Jul. 2008, pp. 1–2. [6] W. H. Kersting and R. C. Dugan, “Recommended practices for distribution system analysis,” in Proc. IEEE PES Power Syst. Conf. Expo., Oct. 2006, pp. 499–504. [7] IEEE PES Distribution Systems Analysis Subcommittee Radial Test Feeders. [Online]. Available: http://ewh.ieee.org/soc/pes/dsacom/ testfeeders.html [8] W. H. Kersting, “A comprehensive distribution test feeder,” in Proc. IEEE PES Transmission Distrib. Conf. Expo., New Orleans, LA, Apr. 2010, pp. 1–4. [9] R. F. Arritt and R. C. Dugan, “The IEEE 8500-node test feeder,” in Proc. IEEE PES Transm. Distrib. Conf. Expo., New Orleans, LA, Apr. 2010, pp. 1–6. [10] S. M. Brahma, “Protecting distribution systems with distributed generation—Are we there yet?” Power Ind. Int., vol. 2, no. I, pp. 61–63, Jun. 2008. [11] J. Nutaro, P. T. Kuruganti, L. Miller, S. Mullen, and M. Shankar, “Integrated hybrid-simulation of electric power and communications systems,” in Proc. IEEE Power Eng. Soc. Gen. Meeting, Jun. 2007, pp. 1–8. [12] H. W. Bindner and O. Gehrke, “System control and communication,” Risø Nat. Lab. Sustainable Energy, Roskilde, Denmark, Risø Energy Rep. 8, The intelligent energy system infrastructure for the future, pp. 39– 42, 2009. [13] M. Baran and T. E. McDermott, “Distribution system state estimation using AMI data,” in Proc. IEEE PES PSCE, Seattle, WA, Mar. 2009, pp. 1–3. [14] CIM User Site. [Online]. Available: http://www.ucaig.org [15] MultiSpeak. [Online]. Available: http://www.multispeak.org Robert F. Arritt (M’96) received the B.S.E.E. degree from West Virginia Institute of Technology, Montgomery, in 2000, and the M.S.E.E. degree from Worcester Polytechnic Institute, Worcester, MA, in 2005. He is currently a Power Systems Engineer with the Electric Power Research Institute (EPRI), Knoxville, TN. His employment experience included Raytheon, Sudbury, MA, where he worked in the Power and Electronic Systems Department. He has spent most of his career designing and modeling power systems from the electronics level to ac generation. Recently, he has been actively involved in distributed generation impact studies and the EPRI Green Circuits effort. Mr. Arritt was the recipient of the 2006 Raytheon Technical Honors Award for Peer and Leadership Recognition for Outstanding Individual Technical Contribution and the 2005 Raytheon Author’s Award for work on phase-shifted transformers for harmonic reduction. Roger C. Dugan (M’74–SM’81–F’00) received the B.S.E.E. degree from Ohio University, Athens, in 1972, and the M.Eng. degree from Rensselaer Polytechnic Institute, Troy, NY, in 1973. From 1992 to 2004, he was a Senior Consultant with Electrotek Concepts, Knoxville, TN. From 1973 to 1992, he held various positions in the Systems Engineering Department, Cooper Power Systems, in Canonsburg, PA, and Franksville, WI. He is currently a Senior Technical Executive with the Electric Power Research Institute, Knoxville. He has worked on many diverse aspects of power engineering over his career because of his interests in applying computer methods to power system simulation. The focus of his career has been on utility distribution systems. He is the coauthor of Electrical Power Systems Quality (McGraw-Hill, 2003). Mr. Dugan was the recipient of the 2005 IEEE Excellence in Distribution Engineering Award. He is the Chair of the Test Feeder Working Group of the Distribution System Analysis Subcommittee of the IEEE Power and Energy Society Power Systems Analysis, Computing, and Economics Committee.