Impacts of Voltage Control Methods on Distribution Circuit s Photovoltaic (PV) Integration Limits

inventions Article Impacts Voltage Methods on Distribution Circuit s Photovoltaic () Integration Limits Anamika Dubey School Electrical Engineering Computer Science, Washington State University; Pullman, WA 99164, USA; anamika.dubey@wsu.edu; Tel.: +1-509-335-1865 Received: 25 September 2017; Accepted: 18 October 2017; Published: 25 October 2017 Abstract: widespread integration photovoltaic () units may result in a number operational issues for utility distribution system. advances in smart-grid technologies with better communication capabilities may help to mitigate se challenges. objective this paper is to evaluate multiple voltage methods compare ir effectiveness in mitigating impacts high levels penetrations on distribution system voltages. A Monte Carlo based stochastic analysis framework is used to evaluate impacts integration, with without voltage. Both snapshot flow time-series analysis are conducted for feeder with varying levels penetrations. methods are compared for ir impacts on (1) feeder s hosting capacity; (2) number voltage violations magnitude largest bus voltage; (3) net reactive dem from substation; (4) number switching operations feeder s legacy voltage support devices i.e., banks load tap changers (s). simulation results show that voltage help in mitigating overvoltage concerns increasing feeder s hosting capacity. Although, legacy solves voltage concerns for primary feeders, a smart inverter is required to mitigate both primary secondary feeder voltage regulation issues. smart inverter, however, increases feeder s reactive dem number switching operations. For 34.5-kV test circuit, it is observed that reactive dem increases from 0 to 6.8 MVAR on enabling Volt-VAR for inverters. total number operations over a 1-year period also increases from 455 operations to 1991 operations with Volt-VAR mode. It is also demonstrated that by simply changing mode banks, a significant reduction in unnecessary switching operations for banks is observed. Keywords: distributed energy resources (DERs); solar energy; photovoltaic systems (s); voltage regulation; voltage ; smart inverters 1. Introduction With incentivized rapid decarbonization electric generation industry aggressive renewable portfolio stards (RPS) in most states, an aggressive integration distributed energy resources (DERs) largely in distribution system is expected [1,2]. To this regard additionally owing to improvements in technology performance, decrease in cost, growing consumer interest, grid-tied photovoltaic () generation has been increasing worldwide. International Energy Agency (IEA) estimates that by 2050, generation will contribute to 16% world s electricity out which 20% total capacity will come from residential installations [3]. Since utility distribution systems are designed for a centralized generation are optimized for unidirectional flow, integration high levels penetration may disrupt system s normal conditions [4,5]. In literature, several studies have been conducted to determine impacts systems on distribution circuit operations [6 11]. Multiple operational issues can result from integrating Inventions 2017, 2, 28; doi:10.3390/inventions2040028 www.mdpi.com/journal/inventions

Inventions 2017, 2, 28 2 20 a large percentage distributed s including feeder overvoltage problems, increased voltage variability [6 9], rmal limit violations, increased feeder loses [10], an increase in number switching regulator tap operations [11]. increasing rate integration undesirable impacts s on distribution circuit call for determining largest capacity a given distribution circuit can accommodate without violating circuit s operational limits. obtained capacity is referred to as circuit s hosting capacity or accommodation limit. emerging DER integration trend also calls for methods to mitigate concerns facilitate integration. In literature, several methods have been proposed to determine feeder s maximum penetration limits [12 26]. se methods are briefly discussed in Section 1.1. For distribution feeders, voltage problems particularly overvoltage issues are primary concern are ten limiting criteria for integrating additional s [24]. existing methods do not include voltage devices while evaluating feeder s integration limits. Since voltage smart inverter is becoming an industry norm; it is imperative to revise hosting capacity study include voltage techniques in analysis framework. Furrmore, along with feeder s hosting capacity, or key parameters affecting feeders with high penetrations should be explored. objective this paper is to evaluate impacts residential deployments on feeder customer voltages while including voltage methods into analysis framework. primary research questions that this article aims at addressing are following: What are impacts different voltage modes for both distribution system s legacy smart inverters on feeder s hosting capacity, feeder s primary secondary voltage violations, feeder s reactive dem? What are impacts smart inverter with high levels penetration on feeder s legacy devices how some undesirable impacts can be mitigated? To this regard, multiple voltage methods are implemented. impacts deployment when voltage is included are assessed using customer- system-level metrics. voltage methods are compared for ir effectiveness in mitigating impacts high levels distributed deployment. Furrmore, a simple case study is presented to mitigate high number tap operations resulting from smart inverter. 1.1. Literature Review undesirable impacts high percentages integration call for study to determine circuit s hosting capacity [13 23]. Given a large number potential deployment scenarios, researchers have primarily looked into simulation-based frameworks [13 22]. Most se methods eir used a simplified feeder model [13 15] or simulated systems at only at a few specific locations [13 15]. Reference [16,17] proposed a stochastic analysis framework to obtain integration limits by simulating large numbers potential deployment scenarios by varying residential panel locations sizes. Building upon framework proposed in [17], a stochastic analysis approach was proposed [20] while taking hourly variations in load dem generation into consideration. Although integration a high percentage may result in multiple operational challenges for distribution system, feeder overvoltage problem is one major concerns. Given impacts DERs on feeder voltages, researchers have worked actively on optimizing feeder voltages reactive (Volt-VAR optimization) by ling coordinating distribution system s legacy devices ( banks load tap changers (s)) new devices (smart inverters) [27 30]. existing algorithms use centralized or distributed approach to optimize real-time operation feeder, while assuming that an advanced communication system exists to coordinate feeder devices. For example, in [27] authors propose a non-cooperative game for Volt-VAR optimization (VVO) while achieving an optimal energy consumption. In [28], a mixed-integer quadratically constrained programming (MIQCP) problem is formulated to optimize

Inventions 2017, 2, 28 3 20 VVO by ling coordinating banks, voltage regulators, load-tap changers for day-ahead operation. References [29,30] present a discussion on multiple Volt-VAR methods for a utility distribution feeder. Unfortunately, existing hosting methods freeze voltage devices while calculating hosting capacity [17,19 21]. rationale for assumption being legacy voltage devices such as s banks are not fast enough,, refore, momentary overvoltage condition will be observed. Additionally, smart inverter for units is also disabled. primary objective calculating hosting capacity is to inform utility providers about limits integration capacity for ir feeder without necessitating grid upgrades [22]. refore, given recent advances in voltage techniques an increased deployment s with smart inverters, including voltage in hosting analysis is necessary. One relatively recent works [31] aims at addressing above problem. article evaluates impacts Volt-VAR smart inverters on hosting capacity [31]. However, authors neir consider or voltage methods/devices nor evaluate impacts voltage on or system parameters. It should be noted that voltage may improve hosting capacity but result in undesirable system impacts, including high reactive dem from substation or an increased number switching operations. This calls for a more detailed analysis comparison to multiple voltage methods/devices for ir pros cons. voltage can be achieved using: (1) feeder s legacy devices such as s banks; (2) new voltage regulation devices such as smart inverters connected to DERs. In this work, we present a thorough evaluation different voltage methods applied to both legacy new devices. analysis, refore, helps to evaluate compare voltage methods devices for ir effectiveness in improving hosting capacity for ir impacts on distribution system technical constraints. 1.2. Objectives Assumptions This paper aims at evaluating impacts high percentages integration on distribution system overvoltage conditions while taking voltage approach into consideration. It should be noted that this paper has a different objective when compared to literature on feeder voltage methods [27 30]. Our objective is to evaluate relative benefits legacy new voltage devices by comparing ir different modes. voltage methods are compared for ir effectiveness in mitigating integration challenges, in facilitating integration additional s, for ir impacts on or system parameters. analysis framework assumptions are summarized next. Voltage Framework following methods are included in this analysis: (1) ling only legacy devices banks s; (2) ling only smart inverters factor Volt-VAR ; (3) ling both legacy devices smart inverters. Metrics to Quantify Impacts following metrics are used to quantify impacts voltage approach: (1) feeder s hosting capacity for primary bus overvoltage; (2) number customers with voltage violations; (3) largest primary secondary bus voltages; (4) reactive supplied by substation; (5) number change operations. Analysis Approach presented analysis framework includes both snapshot time-series flow studies. snapshot flow analysis is conducted at a specific point, in this case, at feeder s minimum load condition. time-series analysis is simulated for 1-year using 1-year load dem data available in 1-h interval 1-year generation data available in 1-min interval. 1.3. Contributions This paper extends existing models for hosting capacity study by including feeder voltage into analysis framework. Essentially, this paper presents a study on (1) impacts voltage

Inventions 2017, 2, 28 4 20 in mitigating overvoltage problem improving feeder s integration limits; (2) or parameters that should be compared to evaluate effectiveness a voltage approach in solving overvoltage problem; (3) impacts voltage on or system parameters, including feeder s reactive dem switching legacy voltage support devices. Given push towards changing stards for DER integration, this paper provides a useful study for assessing best approaches to solve integration challenges. distribution circuit selected for this analysis is adapted from EPRI s 34.5-kV test circuit [32]. A background on hosting capacity problems voltage methods are provided in Section 2. impact analysis hosting capacity framework along with data modeling requirement to conduct study is detailed in Section 3 followed by verification method results in Section 4. A detailed discussion simulation results is presented in Section 5, followed by conclusions in Section 6. 2. Photovoltaic () Hosting Capacity Problem Voltage Framework A feeder s hosting capacity is defined as largest generation that can be accommodated without violating circuit s operational limits. This study is concerned with overvoltage recorded in primary wires due to integration. hosting capacity is equal to largest, for which feeder voltages are within acceptable ANSI voltage limits i.e., <1.05 pu [24]. In order to capture stochasticity deployment resulting from uncertainties in locations size, we use Monte Carlo simulation approach [17]. This section presents a discussion on hosting problem voltage methods. 2.1. Monte Carlo Based Method for Hosting Problem Feeder s actual hosting capacity can be obtained only by simulating all possible deployment scenarios. Since a large number deployment scenarios are possible by varying location size panels, hosting analysis is done using Monte Carlo approach [17]. Monte Carlo approach is used to simulate a large but finite number deployment scenarios in a systematic way [20]. This section details a mamatical framework to quantify feeder s hosting capacity. following terms are defined in order to formulate hosting problem. Customer Penetration Level (C i pen) Defined as percentage customers equipped with panels. C i pen is obtained by populating i% customers with panels, where 0 i 100. Penetration Level ( i pen) Defined as total generation in kw added to distribution circuit corresponding to ith customer penetration level (C i pen). Deployment Scenarios (x i j ) A deployment scenario characterizes locations sizes individual panels at each customer load location for a given customer penetration level. Here, a deployment scenario, x i j, represents jth deployment scenario corresponding to ith customer penetration level. hosting problem is defined as follows. For a given distribution circuit, let k number deployment scenarios are simulated at each customer penetration level, each representing one Mote Carlo run. Since in this study, k Monte Carlo runs/scenarios are simulated at each customer penetration level, method is termed as k-run Monte Carlo study (k-run MCS). feeder s hosting capacity is defined in terms First- All-hosting capacity as following: First-hosting Capacity (H 1,k ) First-hosting capacity is equal to lowest penetration for which at least one scenario in a k-run MCS observes an overvoltage condition. All-hosting capacity (H 100,k ) All-hosting capacity is defined as lowest penetration such that all k scenarios (i.e., 100%) in a k-run MCS observe an overvoltage condition.

Inventions 2017, 2, 28 5 20 First-hosting capacity (H 1,k ) All-hosting capacity (H 100,k ) are obtained by solving (1) (2). { H 1,k = min pen P(V i max,k i > 1.05) 1 } (1) i S k { ) (Vmax,k i > 1.05 H 100,k = min i S i pen P } = 1 where, S be set discrete customer penetration levels indexed by i, S {1, 2,..., i,... 100} be set all penetration levels indexed by customer penetration level pen i, {pen, 1 pen, 2..., pen, i... 100 V i max,k pen } Set maximum primary ( voltages recorded ) for k deployment scenarios simulated at ), V max x2 i }. C i pen, {V max (x i 1 2.2. Voltage Framework ),..., V max ( x i k With ongoing emphasis on grid-modernization enabling new communication capabilities, a realistic impact analysis framework must include voltage capabilities. Depending upon level grid modernization type systems deployed, feeder voltages can be potentially regulated by using legacy voltage devices new equipment. Traditionally, distribution systems include voltage regulators, s, banks for voltage support. With advancement in technology ongoing efforts for new DER interconnection stards, s with smart inverters are expected to be installed. A smart inverter can provide reactive support for feeder by in several modes. se are relatively faster than legacy devices thus potential cidates for mitigating voltage quality problems resulting from generation variability. In this paper, operation both legacy voltage devices new equipment is included. methods evaluated compared in this paper are detailed in this section also summarized in Table 1. Table 1. Voltage methods. (2) Method Device Benefits Disadvantages Required Communication 1 2 3 Legacy devices only s *, voltage regulators, banks New device only * smart inverters Both legacy new devices Require no additional investments Does not increase feeder s reactive dem Fast response can mitigate variability concerns Coordinated can better optimize voltages reactive dem Slow in operation due to time-delayed response Increased operational cost due to more frequent switching Added investment Increases feeder s reactive dem May requires communication optimization framework to achieve grid objectives Can be based on local system variables or external SCADA * signals Can be pre-programmed (PF * ) or require real-time internal communication between output panel voltage at PCC * * Load tap changer transformer, SCADA Supervisory data acquisition, Photovoltaic, PF factor, PCC Point common coupling. 2.2.1. 1 Voltage with Legacy Devices Distribution circuits are equipped with voltage devices such as s, voltage regulators, banks for voltage support. existing integration studies freeze operation legacy devices while calculating feeder s hosting capacity. In this method, operation legacy devices is included while calculating hosting capacity. It should be noted that re is a delay in operation legacy devices. Thus, feeder may record an overvoltage condition for a few minutes until device operates. devices operate based on local information settings, thus do not require an external communication interface.

Inventions 2017, 2, 28 6 20 Inventions 2017, 2, 28 6 20 2.2.2. 2 Voltage with Smart Inverters 2.2.2. 2 Voltage with Smart Inverters following two voltage regulation methods are implemented for smart inverters: (1) Fixed factor following ; two voltage (2) Volt-VAR regulation. methods Note that are implemented panels can for also smart regulate inverters: feeder (1) voltages Fixed by ling factor ir ; active (2) Volt-VAR generation. eir by Note setting that up a generation panels can limit also or regulate using Volt-Watt feeder voltages methods. by ling However, ir active active generation methods eir are by likely setting to be up not a generation preferred by limit or owners. using refore, Volt-Watt in this study, methods. only However, methods active based on reactive methods are likely are included. to be not preferred by owners. refore, in this study, only methods based on reactive are included. Fixed Power Factor Usually, panels operate at a unity factor only Fixed generating Power active Factor. Usually, smart inverter connecting panels operate at a panel unity to grid factor can only be generating programmed active to allow. panels smart to operate inverter at a lagging connecting factor panel absorb to reactive grid can be to programmed mitigate overvoltage to allow concern. panels to impacts operate at a lagging factor factor panels absorb on reactive distribution to circuit mitigate are evaluated overvoltage by programming concern. smart impacts inverters to operate factor at a lagging panels factor. on Two distribution cases are simulated: circuit are (1) evaluated at 0.995 lagging by programming factor; smart inverters (2) at 0.98 to operate lagging at a lagging factor. This factor. Two method cases can are be simulated: simply implemented (1) at 0.995 lagging by one-time programming factor; (2) at 0.98 inverters, lagging thus not factor. requiring This any critical method communication can be during simply operation. implemented by one-time programming inverters, thus not requiring any critical communication during operation. Volt-VAR This function allows on reactive output s according Volt-VAR This function allows on reactive output s to voltage at point connection system available reactive. In this according to voltage at point connection system available reactive study, reactive generated or absorbed by smart inverter for Volt-VAR. In this study, reactive generated or absorbed by smart inverter for follows curve shown in Figure 1. available reactive (Q available ) is calculated using Volt-VAR follows curve shown in Figure 1. available reactive ( (3). In this approach, inverters operate based on voltages at ir point common ) is calculated using (3). In this approach, inverters operate based on voltages at coupling (PCC). ir point common coupling (PCC). Q available = S 2 P2 (3) where, = (3) where, S is apparent rating smart inverter connected to. is P is apparent current active rating generation smart inverter connected panel. to. is current active generation panel. Figure1. 1. Volt-VAR Curve followed by smart inverters to VAR output s. 2.2.3. 3 both Legacy Devices Smart Inverters This approach enables both legacy devices smart inverters. s, banks, voltage regulators, smart inverters operate based on only local information settings. smart inverters eir work based on one-time programming ( factor ) or local bus information (Volt-VAR ). Since in this analysis, both legacy devices smart inverters operate based on ir respective local settings with no coordination, no external communication is required. 3. Analysis Tools Proposed Methodology This section presents a a discussion on on analysis analysis tools tools required required for for study, study, selected selected test circuit, test circuit, proposed proposed methodology. methodology. We implement We implement analysis analysis framework framework using OpenDSS using OpenDSS [33] [33] MATLAB. analysis is conducted using a test circuit adapted from 34.5-kV EPRI test feeder [32] available open source for distribution system analysis.

Inventions 2017, 2, 28 7 20 Inventions MATLAB. 2017, 2, 28 analysis is conducted using a test circuit adapted from 34.5-kV EPRI test feeder 7 [32] 20 available open source for distribution system analysis. 3.1. Analysis Tools 3.1. Analysis proposed Tools framework requires a detailed distribution feeder model an interface to simulate proposed analyze framework multiple requires case studies. a detailed A detailed distribution feeder feeder model model starting anfrom interface to substation simulate down analyze to individual multiple customer case studies. locations A detailed is simulated feeder in OpenDSS model starting [33]. fromrepresentative substation models downfor to customer individual loads customer are also locations developed is simulated using insubstation OpenDSS [33]. load data representative customer models load characteristics. for customer loads are units alsoare developed modeled using equipment substation load model data for systems customer that load are characteristics. available in OpenDSS. MATLAB units are modeled is used to using simulate equipment multiple model deployment for systems scenarios. that are available impact in OpenDSS. voltage MATLAB studies is used to are simulated multiple by interfacing deployment MATLAB scenarios. with OpenDSS. impact voltage studies are simulated by interfacing MATLAB with OpenDSS. 3.1.1. Distribution Circuit Simulator OpenDSS 3.1.1. Distribution Circuit Simulator OpenDSS OpenDSS is a comprehensive electrical system simulation tool for electric utility distribution system OpenDSS supporting is a comprehensive all frequency electrical domain system analyses. simulation It supports tool for electric quasi-static utility distribution unbalanced three-phase system supporting all flow frequency analysis domain for radial analyses. Itlooped supports networks quasi-static while including unbalanceduser-defined three-phase flow algorithms analysis for for shunt radial s, looped voltage networks regulators, while including s, user-defined smart-inverters. algorithms simulator also for shunt facilitates s, DER integration voltage regulators, analysis s, provides smart-inverters. pre-defined models simulator DERs also facilitates ir DER blocks. integration OpenDSS analysis can be implemented provides pre-defined as both a models st-alone DERs executable ir program blocks. an in-process OpenDSS Component can be implemented Object Model as both (COM) a st-alone server DLL executable that is designed program to be riven in-process by a Component variety existing Object stware Model (COM) platforms. server In DLL this study, that is OpenDSS designed to is executed be drivenusing by a MATLAB variety program. existing stware platforms. In this study, OpenDSS is executed using MATLAB program. 3.1.2. Photovoltaic System Model 3.1.2. Photovoltaic System Model An in-built model for systems is available in OpenDSS. model is capable simulation studies Anin in-built time steps model greater for than systems or equal is available to 1 s. in OpenDSS. system is modeled is capable as a simulation delivery object studiesproducing time steps greater according than or to equal some to 1generation s. function. system is A modeled simplified as a block diagram delivery object producing system model according with inverter to some generation is illustrated function. in Figure A simplified 2. active block diagram, P, is a function system model irradiance, with inverter temperature, is illustrated in Figureat 2. active mpp at, a selected P, is a function temperature irradiance, at an irradiance temperature, 1.0 kw/m rated 2. reactive at mpp at a selected for system temperature is specified at an separately irradiance from 1.0 kw/m active 2. reactive eir as fixed for kvar system values, is specified or as a separately fixed from factor active value. Additionally, eir as fixed inverter kvar values, or is capable as a fixed simulating factorvolt-var, value. Additionally, Volt-Watt, dynamic inverter reactive current, is capable user-defined simulating Volt-Var, modes. Volt-Watt, dynamic reactive current, user-defined modes. Smart Inverter Volt-Var, Volt-Watt, Userdefined models, etc. Irradiance Temperature Power... System Model Grid Active (P) Reactive (Q) Figure2. 2. Simplified block diagram OpenDSS model for for system [33]. 3.1.3. Required Circuit Data 3.1.3. Required Circuit Data detailed feeder model including circuit data, load data, DER data is required to conduct A detailed feeder model including circuit data, load data, DER data is required to conduct a realistic distribution system analysis. realistic distribution system analysis. Circuit data Includes one-line diagram feeder, substation model, distribution line characteristics Circuit data Includes including impedance, one-line diagram ampacity, voltage feeder, levels, substation connections, model, distribution details line regarding characteristics distribution including transformers impedance, ampacity, banks. voltage levels, connections, details Load regarding data Includes distribution transformers hourly load dem data banks. measured at substation for one year or more as well as characteristics individual customer loads distributed across feeder.

Inventions 2017, 2, 28 8 20 Inventions Load 2017, data Includes 2, 28 hourly load dem data measured at substation for one year 8 or 20 more as well as characteristics individual customer loads distributed across feeder. DER data Includes details existing DER/ integration. necessary data include panels location, size, generation priles. High-resolution generation data data for for one one year year or or for for a few a few representative days days should be be provided. 3.2. Selected Distribution Circuit Test Circuit analysis is conducted using 34.5-kV EPRI test feeder [32]. substation transformer supplying selected feeder has a anominal MVA rating 45 45MVA. substation transformer is supplying is for for two tweeders, one one which is isrepresented as asan equivalent load connected to substation with aa rated rated load load dem dem 23.5 23.5 MVA. MVA. detailed detailed distribution distribution circuit model circuit is model available is available for or for feeder or (Feeder feeder 1), (Feeder with 1), with one-line diagram one-line diagram shown inshown Figurein 3Figure plotted3 using plotted toolbox using provided toolbox provided by Reference by Reference [34]. [34]. integration integration analysis analysis simulated is simulated for Feeder for 1. Feeder For 1. selected For selected distribution distribution circuit, circuit, 1-year customer 1-year customer load prile load data prile is available data is available at 1-h resolution. at 1-h resolution. This circuit This circuit has three has three banks banks ratings ratings 0.9 MVAR 0.9 MVAR ( ( 1), 1.2 MVAR 1), 1.2 MVAR ( ( 2), 2), 1.2 MVAR 1.2 ( MVAR ( 3). 3). banks 1 banks 2 are 1 2 are kvar in kvar mode, while mode, while 3 is always 3 online. is always online. substation transformer substation transformer is equippedis with equipped an. with an. 3 1 2 Figure 3. Detailed one-line diagram test circuit (Feeder 1) [32]. For given feeder, flow analysis is conducted for year. peak load dem For given feeder, a flow analysis is conducted for 1 year. peak load dem recorded at substation for year is equal to 52.15 MVA (Feeder Feeder 2). For snapshot recorded at substation for year is equal to 52.15 MVA (Feeder 1 + Feeder 2). For snapshot flow analysis, a minimum load condition for feeder is identified. In this study, flow analysis, a minimum load condition for feeder is identified. In this study, minimum minimum load is assumed to be 20% feeder s peak load dem is equal to 10.43 MVA. load is assumed to be 20% feeder s peak load dem is equal to 10.43 MVA. For For time-series analysis, 1-year load dem data at an hourly resolution 1-year prile time-series analysis, 1-year load dem data at an hourly resolution 1-year prile data at data at 1-min resolution is utilized [35]. prile in 1-min resolution for 1-day, shown in 1-min resolution is utilized [35]. prile in 1-min resolution for 1-day, shown in Figure 4. Figure 4. same prile is replicated for each day year to obtain 1-year prile. same prile is replicated for each day year to obtain 1-year prile.

Inventions Inventions 2017, 2017, 2, 28 2, 28 9 9 20 20 Solar Irradiance (W/m 2 ) Figure Figure 4. 4. generation prile prile for for one one day day at 1-min at 1-min resolution [35]. [35]. 3.3. 3.3. Hosting Impact Analysis Methodology This This section presents a summary a stochastic analysis framework used used for for determining impacts systems on on distribution circuit. approach includes simulating multiple deployment scenarios, quantifying quantifying impacts impacts using snapshot using snapshot time-series time-series analysis, analysis, evaluating customer evaluating customer system-level impact system-level metrics. impact metrics. 3.3.1. 3.3.1. Simulate Deployment Scenarios deployment deployment scenarios scenarios are simulated are simulated using Monte using Carlo Monte approach Carlo approach by associating by associating a uniform a rom uniform variation rom tovariation locations to [17] locations by[17] identifying by identifying realistic realistic system sizesystem using [35]. size using At each [35]. customer At each penetration customer penetration level (C i pen), level we simulate ( ), we 100simulate unique 100 deployment unique scenarios deployment (Xk i ). scenarios method ( ). to systematically method to simulate systematically jth simulate deployment jth scenario, deployment where j scenario, = {1, 2,. where.., 100} is = as {1,2, follows.,100} First, is as a 2% follows. distribution First, a 2% customers distribution (C 2 pen) arecustomers integrated( with ) s. are integrated customer with penetration s. level customer is incremented penetration inlevel a stepis incremented 2% until it reached in a step 100% 2% additional until reached s are 100% deployed. additional This constitutes s are one deployed. deployment This constitutes scenario. one above deployment process isscenario. repeated 100above timesprocess to obtainis 100 repeated deployment 100 times to scenarios. obtain 100 Note that deployment above process scenarios. results Note in athat total above 5000 process deployment results cases, in a with total 100 unique 5000 deployments corresponding cases, with 100 tounique each customer deployments penetration corresponding level. Please to each refer customer [17,20] penetration for details level. regarding Please refer simulations. to [17,20] for details regarding simulations. 3.3.2. 3.3.2. Quantify Quantify Impacts Impacts on on System System Voltage Voltage impacts impacts integration integration are evaluated are evaluated using snapshot using snapshot time-series time-series flow analysis. flow analysis. Snapshot Power Flow Analysis For each deployment scenario, a three-phase unbalanced load Snapshot flow analysis Power Flow is simulated Analysis For at feeder s each minimum deployment load scenario, condition. a three-phase voltage unbalanced methods load flow areanalysis implemented is simulated load at flow feeder s analysisminimum simulated load again condition. for each voltage deployment scenario. methods are snapshot implemented flow load analysis flow analysis helps inis identifying simulated again impacts for each increasing deployment penetrations scenario. onsnapshot feeder s primary flow secondary analysis helps voltages in identifying feeder s reactive impacts increasing dem. Time-series penetrations Analysis For on feeder s primary each voltage secondary approach, voltages time-series feeder s reactive analysis for 1-year dem. is simulated Time-series using Analysis For 1-min quasi-static each voltage flow simulations. approach, A time-series analysis is conducted for 1-year is after simulated implementing using 1-min different quasi-static voltage flow methods. simulations. A A deployment time-series analysis scenariois resulting conducted inafter overvoltage implementing condition different when voltage methods. A are not deployment employed scenario is selected resulting for in analysis. overvoltage Using condition time-series when voltage analysis, number methods are tap not changes employed is selected bank for switching analysis. operations Using are time-series recorded. analysis, time-series number analysis helps in tap understing changes impacts bank integration switching operations voltageare recorded. on feeder s time-series legacy devices. analysis helps in understing impacts integration voltage on feeder s legacy devices.

Inventions 2017, 2, 28 10 20 3.3.3. Calculate Hosting Capacity Quantify Voltage Impacts Based on load flow analysis, feeder s integration limits corresponding to First-hosting All-hosting capacities are calculated. Along with feeder s hosting capacity, following additional parameters are measured to compare different voltage methods. Impacts on Secondary Customer Voltages largest feeder voltage number customers observing overvoltage are also recorded to compare impacts voltage. Impacts on Reactive Power Supplied by Substation For each voltage approach, reactive supplied by substation is recorded compared. Impacts on Voltage Equipment Yearly load flow analysis for base case for each voltage case are simulated total number switching tap changes are recorded compared. 4. Results This section summarizes results impact analysis voltage method for selected distribution feeder. 4.1. Snapshot Power Flow Analysis snapshot flow analysis evaluates impact integration by simulating a three-phase unbalanced load flow at a feeder s minimum load condition equal to 10.43 MVA (20% peak load). Depending upon voltage methods discussed in Section 2, eight separate case studies are simulated. In order to simplify discussion, a summary each case corresponding approach is detailed in Table 2. Table 2. Voltage case studies. Case Study Mode Smart Inverter PF = 0.995 PF = 0.98 Volt-Var Case 1 No Disabled Disabled Disabled Disabled Disabled Case 2 Mode 1 Enabled Enabled Disabled Disabled Disabled Case 3 Mode 2 Disabled Disabled Enabled Disabled Disabled Case 4 Mode 2 Disabled Disabled Disabled Enabled Disabled Case 5 Mode 2 Disabled Disabled Disabled Disabled Enabled Case 6 Mode 3 Enabled Enabled Enabled Disabled Disabled Case 7 Mode 3 Enabled Enabled Disabled Enabled Disabled Case 8 Mode 3 Enabled Enabled Disabled Disabled Enabled 4.1.1. No Voltage (Case 1) In this case, panels are at a unity factor, s are frozen. penetration is increased to 100% amounting to a total 24 MW. hosting capacity number primary secondary buses observing an overvoltage condition are shown in Figure 5. selected distribution circuit can accommodate a 9.3 MW capacity without implementing any voltage method. All-hosting capacity turns out to be 11.4 MW. As can be seen from Figure 5b, 1455, 3990 secondary customers observe overvoltage at First- All-hosting capacity, respectively. Furrmore, as penetration increases, number voltage violation increase sharply with almost all customers recording a voltage violation at 80% penetration.

Inventions 2017, 2, 28 11 20 Inventions 2017, 2, 28 11 20 Inventions 2017, 2, 28 11 20 Maximum Maximum primary primary bus bus voltage voltage (pu) (pu) (b) (b) Figure 5. No voltage case: Largest feeder voltages; (b) Number bus voltage violations. Figure 5. No voltage case: Largest feeder voltages; (b) Number bus voltage violations. Figure 5. No voltage case: Largest feeder voltages; (b) Number bus voltage violations. 4.1.2. Mode 1 Enabling Load Tap Changer () (Case 2) 4.1.2. 4.1.2. Mode Mode 1 Enabling Load Load Tap Tap Changer Changer () () (Case (Case 2) 2) For this case study, legacy devices i.e., s banks is enabled while For conducting For this this case case study, study, impact hosting legacy legacy devices capacity devices i.e., study. i.e., s s Since legacy banks devices banks is take is enabled enabled while some time while to conducting conducting respond; a voltage impact violation impact hosting is expected hosting capacity for capacity study. a short study. Since legacy devices take some time to respond; duration Since until legacy devices activates. take some time results to arespond; voltage violation is expected for a short duration until activates. results are shown are shown a voltage in Figure violation 6. With is legacy expected for enabled, a short duration feeder until records very few activates. cases (5 cases results out inare Figure 6. With legacy enabled, feeder records very few cases (5 cases out 5000) 5000) shown overvoltage in Figure 6. in With primary legacy wire. A enabled, tap change operation feeder records is recorded very few at cases penetration (5 cases out equal overvoltage 5000) in primary wire. A tap change operation is recorded at penetration equal to 4 to 4 overvoltage 16 MW. in secondary primary customers wire. A tap record change multiple operation cases is recorded voltage violations. penetration equal 16to MW. 4 16 secondary MW. secondary customers customers record multiple record cases multiple voltage cases violations. voltage violations. Number Number Buses Buses Maximum Maximum feeder feeder voltage voltage (pu) (pu) Number Number Buses Buses (b) (b) Figure 6. Load tap changers () : Largest feeder voltages; (b) Number Figure Figure bus voltage 6. 6. Load Load violations. tap tap changers changers () () : : Largest Largest feeder feeder voltages; voltages; (b) (b) Number Number bus bus voltage voltage violations. violations. When compared to Case 1, at First- All-hosting capacity limit Case 1 (9.3 11.4 MW When penetration), compared to in Case this case, 1, at 38 First- 104 secondary All-hosting customers capacity report limit an overvoltage Case 1 (9.3 condition, 11.4 When compared to Case 1, at First- All-hosting capacity limit Case 1 (9.3 11.4 MW respectively. penetration), largest in this number case, 38 secondary 104 secondary bus voltage customers violations report for an this overvoltage case is 780, condition, MW penetration), in this case, 38 104 secondary customers report an overvoltage condition, which is respectively. significantly less largest than number number secondary bus voltage bus voltage violations violations recorded for this for case Case is 1. 780, In which Case is respectively. largest number secondary bus voltage violations for this case is 780, which1, significantly approximately less all than 7200 secondary number customers bus record voltage an violations overvoltage recorded at 100% for penetration. Case 1. In Case 1, is significantly less than number bus voltage violations recorded for Case 1. In Case 1, approximately all 7200 secondary customers record an overvoltage at 100% penetration. approximately all 7200 secondary customers record an overvoltage at 100% penetration. 4.1.3. Mode 2 Smart Inverter (Cases 3,4,5) 4.1.3. Mode 2 Smart Inverter (Cases 3,4,5) 4.1.3. This Mode mode 2 Smart enables Inverter smart inverter (Cases 3,4,5) panels. legacy devices is disabled. This smart mode inverters enables are smart programmed inverter for constant panels. factor (PF) legacy devices Volt-VAR is This mode enables smart inverter panels. legacy devices is disabled.. For constant smart inverters factor are programmed, two for cases constant are simulated factor with (PF) all inverters Volt-VAR disabled. smart inverters are programmed for constant factor (PF) Volt-VAR set to. operate For at (1) constant (Case 3) a lagging factor, factor two cases 0.995 are simulated (2) (Case with 4) at all a lagging inverters factor set to. For constant factor, two cases are simulated with all inverters set to operate 0.98. at Volt-VAR (1) (Case 3) a lagging (Case 5) is based factor on 0.995 curve shown (2) (Case in Figure 4) at a 1. lagging factor operate at (1) (Case 3) a lagging factor 0.995 (2) (Case 4) at a lagging factor 0.98. 0.98. For Volt-VAR both constant (Case factor 5) is based Volt-VAR on curve shown modes, in Figure feeder 1. Volt-VAR (Case 5) is based on curve shown in Figure 1. can accommodate a larger For both capacity. constant First-hosting factor is increased Volt-VAR to 17.1 modes, 24 MW (100% feeder customer can accommodate penetration) a For both constant factor Volt-VAR modes, feeder can accommodate a larger larger when smart capacity. inverters are First-hosting a is lagging increased to factor 17.1 0.995 24 MW (100% 0.98, respectively customer penetration) capacity. First-hosting is increased to 17.1 24 MW (100% customer penetration) (see when Figure when 7a,b). smart In Volt-VAR inverters are mode, a lagging feeder can factor accommodate 0.995 100% 0.98, respectively capacity (see without Figure smart inverters are a lagging factor 0.995 0.98, respectively (see Figure 7a,b). 7a,b). violating In Volt-VAR limits for primary mode, overvoltage feeder criteria can accommodate (see Figure 7c). 100% Furrmore, capacity number without violating secondary customers limits for observing primary overvoltage an overvoltage criteria condition (see Figure decreases 7c). Furrmore, for each deployment number secondary customers observing an overvoltage condition decreases for each deployment

Inventions 2017, 2, 28 12 20 In Volt-VAR mode, feeder can accommodate 100% capacity without violating limits for primary overvoltage criteria (see Figure 7c). Furrmore, number secondary customers observing an overvoltage condition decreases for each deployment scenario (Figure 7d f). For Case 3, with smart Inventions inverters 2017, 2, 28 art PF = 0.995, at 100% penetration, a total 100012 secondary 20 customers record an overvoltage (see Figure 7d). When factor is decreased to 0.98 lagging, scenario (Figure 7d f). For Case 3, with smart inverters art PF = 0.995, at 100% numbed penetration, secondary a total customers ( 1000 ) secondary with overvoltage customers record violations an overvoltage furr decreased (see Figure to7d). less When than 20 at customer penetration factor is level decreased C i pento = 0.98 100% lagging, (see Figure numbed 7e). secondary Volt-VARcustomers mode with overvoltage is most effective in regulating violations bothfurr primary decreased secondary to less than feeder 20 at voltages. customer Inpenetration Volt-VARlevel ( ) mode, = 100% (see largest Figure 7e). Volt-VAR mode is most effective in regulating both primary secondary secondary voltage decreases significantly with fewer than eight secondary customers observing any feeder voltages. In Volt-VAR mode, largest secondary voltage decreases significantly overvoltage with at fewer customer than eight penetration secondary level customers (see observing Figure 7f). any overvoltage at customer penetration level (see Figure inverters 7f). in factor Volt-VAR modes will absorb reactive to correct an overvoltage inverters condition. in factor three Volt-VAR methods are modes compared will absorb for ir reactive reactive requirement to correct from an overvoltage substation condition. in Figure 8. three comparison methods is made are atcompared two penetration for ir levels: 40% (corresponding reactive to requirement feeder s from hosting substation capacityin without Figure 8. any ) comparison at is 100%. made at two reactive penetration levels: 40% (corresponding to feeder s hosting capacity without any ) provided by substation increases for each mode as customer penetration (C i pen) is at 100%. reactive provided by substation increases for each mode as customer increased from 40% to penetration ( 100%. Also, reactive drawn from substation increases when ) is increased from 40% to 100%. Also, reactive drawn from substation factor is decreased increases when to 0.98. factor Volt-VAR is decreased method to 0.98. requires Volt-VAR largest method reactive requires largest dem reactive from substation explaining dem its from effectiveness substation in explaining regulatingits secondary effectiveness voltages. in regulating At 100% secondary penetration, voltages. At at PF = 0.995, 100% reactive penetration, at drawn PF = 0.995, is 1.3 MVAR reactive which increases drawn is 1.3 to MVAR 3.8 MVAR which atincreases PF = 0.98 to 3.8 to MVAR at PF = 0.98 to 6.8 MVAR for Volt-Var. 6.8 MVAR for Volt-Var. Power Factor (PF) with PF = 0.995 Case 3 Maximum feeder voltage (pu) Number Buses (d) Power Factor (PF) with PF = 0.98 (lagging) Case 4 (e) (b) Volt-Var Case 5 Maximum feeder voltage (pu) (c) Number Buses (f) Figure 7. Smart inverter : (a c) Largest feeder voltages; (d f) Number bus voltage Figure 7. violations. Smart inverter : (a c) Largest feeder voltages; (d f) Number bus voltage violations.

PF with PF = -0.98 Volt-Var 0 1 2 3 4 5 6 7 8 Reactive supplied (MVAR) Inventions 2017, 2,2017, 28 2, 28 Inventions penetration = 40% Figure Reactive PF 8. with PF = -0.995 penetration = 100% 13 20 13 20 supplied by substation (MVAR) Only smart inverter. PF with PF = -0.98 4.1.4. Mode 3 Enable Both Legacy Devices Smart Inverter (Cases 6, 7 8) Volt-Var Number Buses Maximum feeder voltage (pu) In this mode, both legacy devices smart inverter s are enabled. Three cases 0 1 2 3 4 5 6 7 8 Reactive supplied are simulated by enabling legacy devices with (MVAR) (1) (Case 6) smart inverters at lagging PF = 0.995; (2) (Case 7) smart inverters at lagging PF = 0.98; (3) (Case 8) smart penetration = 40% penetration = 100% inverters in Volt-VAR. ItFigure is observed that with factor results in ainverter better regulation both 8. Reactive supplied by substation (MVAR) Only Figure 8. Reactive supplied by substation (MVAR) Only smartsmart inverter.. primary secondary feeder voltages. When compared to Cases 3 5, enabling 4.1.4. Mode 3 Enable BothLegacy Legacyfor Devices Smart Inverter 6,inverters. 7 6,8)7 increases accommodation limit all modes (Cases smart With 4.1.4. Mode 3 Enable Both Devices Smart Inverter (Cases 8), In when smart mode, inverter is legacy at a lagging PF = 0.995, are feeder is able accommodate this both devices smart inverter s enabled. Threetocases Inare this by mode, both legacy devices smart inverter s are enabled. Three cases enabling legacy with (1) smart inverters at is 100% simulated customer with (see Figure 9a)devices compared to(case 17.1 6) MW (Figure 7a) when are simulated by enabling legacy devices with (1) (Case 6) smart inverters at laggingan PF = 0.995; (2)change (Case 7) is smart inverters at lagging PF =is0.98; to (3)5(Case smart as without disabled. tap observed at generation that equal MW.8)Same lagginginverters PF = 0.995; (2) (Case 7) smart inverters at lagging PF = 0.98; (3) (Case 8) smart in Volt-VAR. enabling, feeder can accommodate 100% customers with when smart It is observed that with factor results in a better regulation both inverters in Volt-VAR. inverters are atfeeder 0.98 voltages. laggingwhen factor, toascases shown inenabling Figure 9b. number primary secondary compared 3 5,in It is observed that with factor results a better regulation both secondary violations limit for magnitude modes largest voltages also decrease increasesvoltage accommodation all secondary smart inverters. With primary secondary feeder voltages. When compared to Cases 3 5, enabling increases, when smart inverter is to at awhen lagging PF = 0.995, isfeeder is ablefigure to accommodate significantly when compared case disabled. 9d shows that for 100% accommodation limit all modes smart inverters. With, 0.995 customer withfor Figure 9a) compared to 17.1 MWcustomers (Figure 7a) when is when Case 6 with lagging pf,(see only a total 62 secondary record overvoltage at 100% smart inverter is at a lagging PF = 0.995, feeder is able to accommodate 100% disabled. An tap change is observed at generation that is equal to 5 MW. Same as without customer penetration level as compared to 1000 customers in Case 3 (see Figure 7d). On reducing enabling, feeder can accommodate customers with when smart customer with (see Figure 9a) compared to 17.1 100% MW (Figure 7a) when is disabled. pf toinverters 0.98 lagging with enabled, number secondary buses recording an are at 0.98 lagging factor, showntoin5 Figure 9b. An tap change is observed at generation that isasequal MW. Same as number withoutenabling overvoltage decrease to 10 cases (seelargest Figure 9e). Furrmore, shown in Figure secondaryfurr less than magnitude secondary voltages alsoas decrease, voltage feeder violations can accommodate 100% customers with when smart inverters are 9c,f,significantly on enabling to with smart inverter in Volt-VAR mode, when compared case when is disabled. Figure 9d shows that for feeder at 0.98 lagging factor, as shown in Figure 9b. number secondary voltage with 0.995 pf, only a total 62bus secondary record 100% doescase not6report anylagging primary or secondary voltagecustomers violations forovervoltage any at deployment violations magnitude largest secondary voltages also decrease significantly customer penetration level as compared customers in Case 3 (see Figure 7d). On reducing when scenarios can accommodate 100% to 1000 penetration. compared lagging case when is disabled. 9d shows that for buses Case 6recording with 0.995 pf toto0.98 with enabled,figure number secondary an lagging decrease customers to less than 10 cases overvoltage (see Figure 9e).atFurrmore, as shown in Figure level as pf, onlyovervoltage a total furr 62 secondary record 100% customer penetration 9c,f, to on 1000 enabling with smart in Volt-VAR feeder with compared customers in Case 3 (seeinverter Figure 7d). On reducing pf tomode, 0.98 lagging does not report any primary or secondary bus voltage violations for any deployment enabled, number secondary buses recording an overvoltage furr decrease to scenarios can accommodate 100% penetration. less than 10 cases (see Figure 9e). Furrmore, as shown in Figure 9c,f, on enabling with smart inverter in Volt-VAR mode, feeder does not report any primary or + PF with PF = 0.995 Case 6 secondary bus voltage violations for any deployment scenarios can accommodate 100% penetration. + PF with PF = 0.995 Case 6 (d) + PF with PF = 0.98 (lagging) Case 7 (d) Figure 9. Cont.

Inventions 2017, 2, 28 Inventions 2017, 2, 28 14 20 14 20 + PF with PF = 0.98 (lagging) Case 7 (e) (b) + Volt-Var Case 8 (f) (c) Figure 9. Cases 6, Legacy 7, 8: Legacy smart : inverter(a c) : (a c) Largest feeder Figure 9. Cases 6, 7, 8: device device smart inverter Largest feeder voltages; voltages; (d f) Number violations. bus voltage violations. (d f) Number bus voltage Time-Series Analysis 4.2.4.2. Time-Series Analysis objective time-series analysis is to evaluate impacts different smart-inverter objective time-series analysis is to (1)(1) evaluate impacts different smart-inverter modes on number tap changes bank switching operations; modes on number tap changes bank switching operations; (2) to(2) to present a simple approach to reduce unnecessary switching legacy devices. time-series present a simple approach to reduce unnecessary switching legacy devices. time-series 𝑖 analysis is conducted for one deployment scenarios with customer penetration (𝐶𝑝𝑒𝑛 ) = i analysis is conducted for one deployment scenarios with customer penetration (C pen ) = 40% 40% that corresponds to capacity hosting capacity for feederenabling without any enabling any voltage that corresponds to hosting obtainedobtained for feeder without voltage.. For yearly analysis, customer load prile in 1-h resolution generation prile For yearly analysis, customer load prile in 1-h resolution generation prile in 1-min in 1-min resolution used. A comparison different for ir tap resolution is used. Ais comparison different modes formodes ir impacts onimpacts tapon changes changes bank bankisswitching presented. Next, a simple approach detailed switching presented.isnext, a simple approach is detailed to is reduce to reduce number switching operations legacy devices. number switching operations legacy devices. 4.2.1. Mode 2 With Only 4.2.1. Mode 2 With Only case, each inverter is at PF 1 i.e., neir generating absorbing In In thisthis case, each inverter is at PF = 1=i.e., neir generating nornor absorbing reactive. is enabled. banks 1 2 are reactive. is enabled. banks 1 2 are in in kvar mode, while bank 3 is a fixed always enabled. positions kvar mode, while bank 3 is a fixed always enabled. positions for fortaps taps are shown for 1in year in Figure 1-year simulation, total statusstatus are shown for 1 year Figure 10. In10. In1-year simulation, a totala 265 tap operations tap operations are recorded. 1 switches 174 times year while 265 are recorded. 1 switches 174 times overover year while 2 2 observes 16 switching operations. observes 16 switching operations.

Inventions 2017, 2, 28 15 20 Inventions 2017, 2, 28 15 20 Inventions 2017, 2, 28 15 20 (b) Figure 10. 10. No for for inverters: Regulator tap operation; (b) (b) bank switching. (b) 4.2.2. 4.2.2. Mode Mode 3 With 3 With Legacy Legacy Smart Smart Inverter Inverter Constant Constant Power Power Factor Factor Figure 10. No for inverters: Regulator tap operation; (b) bank switching. In In this this case, case, inverters inverters are led are led using constant using constant factor mode factor are mode programmed are programmed 4.2.2. to Mode absorb 3 With reactive Legacy. Smart Two Inverter cases are Constant simulated, Power Factor with to absorb reactive. Two cases are simulated, with all inverters all inverters at (1) a lagging PF In at this (1) 0.995 case, a lagging (2) a PF lagging inverters 0.995 are led factor (2) a lagging using 0.98. constant legacy factor devices factor 0.98. operate mode legacy based are on devices ir local operate programmed based settings. on to ir absorb local yearly reactive load. flow settings. Two analysis cases is yearly are simulated, simulated, load flow with analysis number all is simulated, inverters tap changes number at tap (1) a changes lagging PF 0.995 switching (2) a lagging operations factor are recorded. 0.98. legacy devices switching operations are recorded. operate When based inverters on ir local at settings. lagging PF yearly 0.995, load total flow analysis 244 tap is operations simulated, are recorded When inverters at a lagging PF 0.995, a total 244 tap operations are recorded over number year. tap changes switches 231 times switching while operations are recorded. over year. 1 switches 231 times while 2 records records a total total 16 16 switching switching operations. When tap inverters operations decrease at a lagging this PF case, 0.995, however, a total 244 tap operations banks switch are recorded operations. more ten over year. tap operations 1 switches decrease 231 intimes this case, while however, 2 records a total banks switch 16 switching more ten with with factor. On inverters at 0.98 lagging factor, both number operations. factor. tap operations On decrease in this inverters case, however, at 0.98 lagging banks factor, switch both more ten number tap tap operations operations bank bank switching switching increase. increase. With With 0.98 0.98 lagging lagging factor, factor, in in 1-year 1-year a with factor. On inverters at 0.98 lagging factor, both number total total tap 274 274 operations tap tap changes changes are are recorded. recorded. bank switching increase. 1 switches switches With 897 897 0.98 times times lagging while while factor, in 2 1-year switches switches a 72 72 times times total over over 274 year. year. tap changes number number are recorded. tap tap changes changes 1 switches 897 times switching switching while operations operations 2 switches increase increase 72 when when inverters inverters times over are are year. number at at a lower lower tap changes factor. factor. switching operations increase when inverters are at a lower factor. 4.2.3. Mode 3 With Smart Inverter Volt-VAR 4.2.3. Mode 3 With Smart Inverter Volt-VAR Each inverter, in in this case, is is in in Volt-VAR mode that is is actively regulating customer Each voltages. inverter, in this taps case, is in Volt-VAR bank switching mode recorded that is for actively 1-year regulating are shown in in customer voltages. taps bank switching recorded for 1-year are shown in Figure 11. 11. In In this case, number tap tap changes switching operations increase Figure 11. In this case, number tap changes switching operations increase significantly. A total total 556 556 tap tap changes changes is is recorded over a 1-year period. 1 records 1 records a total a total significantly. A total 556 tap changes is recorded over a 1-year period. 1 records a total 708 708 switching 708 switching switching operations, operations, operations, while while while 2 switches 22 switches switches for for 727 for 727 times 727 times times over over over year. year. year. 3, 3, being3, fixed being being fixed fixed is always is is on. always A Volt-VAR on. A Volt-VAR approach approach may approach significantly may may significantly significantly affect affect operation affect legacy operation operation devices. legacy legacy devices. Figure 11. Cont. (b) (b)

Inventions 2017, 2, 28 16 20 Inventions 2017, 2, 28 16 20 (c) (d) Figure 11. Smart inverter Volt-VAR : Regulator Tap Operation; (b) bank 1 Figure 11. Smart inverter Volt-VAR : Regulator Tap Operation; (b) bank 1 switching; (c) bank 2 switching;, (d) bank 3 switching. switching; (c) bank 2 switching;, (d) bank 3 switching. 4.2.4. Coordinated Smart Inverter Volt-VAR with Modified Settings 4.2.4. Coordinated Smart Inverter Volt-VAR with Modified Settings Based on previous three cases, we observe that on implementing reactive Based methods onfor s, previous number three cases, regulator we tap observe operations that on implementing bank reactive switching increases methods significantly. for s, This will number lead to an regulator increased tap operations cost unnecessary aging bank switching equipment. increases significantly. To reduce This will impact lead to Volt-VAR an increased operation on legacy cost unnecessary devices, we implement aging modified equipment. settings for banks. banks 2 3, previously in kvar To reduce impact Volt-VAR operation on legacy devices, we implement modified mode are reconfigured to operate in voltage mode. Recall that Volt-VAR settings for banks. banks 2 3, previously in kvar mode supplies required reactive dem while regulating feeder voltages. mode are reconfigured banks, refore, to operate undergo in voltage unnecessary operations mode. while Recall compensating that Volt-VAR for time-varying mode supplies reactive required dem/generation reactive dem resulting while from regulating fast Volt-VAR feeder. voltages. This unnecessary banks, switching refore, operation undergois unnecessary avoided when operations while banks compensating are reconfigured for to time-varying work in voltage reactive dem/generation mode. resulting from fast Volt-VAR. This unnecessary switching operation is avoided when During 1-year banks simulation are reconfigured duration, with to work modified in voltage, mode. banks 1 2 are During not required 1-year operate. simulation In beginning duration, with simulation, modified, banks 1 2 banks are switched 1 2f are not required operate. are not enabled In for beginning entire 1-year duration. simulation, This is expected banks as 1smart 2inverters are switched f in are Volt-VAR mode are providing required reactive support to grid, thus not not enabled for entire 1-year duration. This is expected as smart inverters in Volt-VAR requiring any operation. 3 being a fixed bank remains enabled for mode are providing required reactive support to grid, thus not requiring any entire 1-year duration. total number regulator tap operations with modified bank operation. s reduces to 468 3 tap being changes a fixed from 556 tap operations bank remains as recorded enabled in for Section entire 4.2.3. 1-year duration. With new total number regulator mode, no tap violations operations in feeder with voltage modified are recorded eir for bank primary s reduces feeders to 468 or for tapsecondary changesservice from 556 wires. tap operations as recorded in Section 4.2.3. With new This study mode, demonstrates no violations that a coordinated feeder voltage operation are recorded legacy devices eir for new primary devices feeders can be or for secondary enabled service by using wires. simple mode change. A more complex coordination approach can also be This developed studythat demonstrates may require that communication a coordinated operation direct legacy devices legacy new new devices devices. can be enabled However, by using proposing simple such mode scheme change. is not A within more complex scope coordination this paper. approach can also be developed that may require communication direct legacy new devices. However, 5. Discussion proposing such scheme is not within scope this paper. A discussion on several voltage methods ir impacts on feeder s hosting 5. Discussion capacity on or feeder characteristics is detailed in this section. Table 3 presents a detailed comparison eight case studies. A discussion on several voltage methods ir impacts on feeder s hosting capacity on or feeder characteristics is detailed in this section. Table 3 presents a detailed comparison eight case studies.

Inventions 2017, 2, 28 17 20 Table 3. Comparison voltage methods. Mode First-Hosting Capacity (MW) Largest Secondary Bus Voltage At Customer Penetration Level (C i pen) = 40% Number Secondary Violations Reactive Power Supplied by Substation Number Tap Operations Number Switching Case 1 No 9.3 1.07 pu 1192 0 * NA NA Case 2 Mode 1 24 (100%) 1.063 pu 47 0 * 265 190 Case 3 Mode 2 17.1 (71.25%) 1.063 pu 87 0 * NA NA Case 4 Mode 2 24 (100%) 1.057 pu 8 0.08 MVAR NA NA Case 5 Mode 2 24 (100%) 1.051 pu 2 2.29 MVAR NA NA Case 6 Mode 3 24 (100%) 1.057 pu 6 0 * 244 247 Case 7 Mode 3 24 (100%) 1.057 pu 7 0.1 MVAR 274 969 Case 8 Mode 3 24 (100%) 1.041 pu 0 2.54 MVAR 556 1435 * Reactive is injected back to substation. 5.1. Hosting Capacity for Primary Feeder Overvoltage Concern Feeder s hosting capacity increases on implementing voltage methods. For selected feeder, 100% customers could be integrated with by simply including existing in analysis. Smart inverter is also effective in increasing feeder s hosting capacity. When smart inverters are eir at 0.98 lagging factor or in Volt-VAR mode, feeder can integrate 100% customers with without resulting in any overvoltage condition in primary feeders. With 0.995 lagging factor, integration limit increases to 17.1 MW for First-hosting 24 MW (100%) for All-hosting capacity. Thus, enabling voltage could be an effective way mitigating primary overvoltage concerns due to large percentages residential generation. 5.2. Impacts on Secondary Customer Voltages Along with primary feeder voltages, both legacy device smart inverter helps in regulating secondary customer voltages. Overall, smart inverter is more effective in regulating secondary customer voltages. largest secondary voltages number secondary bus voltage violations decrease on enabling voltage. After decreasing inverter s factor to 0.98 from 0.995, lower secondary bus voltages a lesser number voltage violations are recorded. Volt-VAR smart inverter results in best regulation for secondary bus voltages with least number overvoltage cases. For selected feeder, Volt-VAR along with enables 100% integration without resulting in any cases primary or secondary voltage violations. 5.3. Reactive Power Supplied by Substation does not increase reactive dem. However, smart inverter comes at cost increasing reactive dem from substation. As smart inverter operates at lower lagging factor, reactive dem increases. Volt-VAR results in largest increase in reactive requirement from substation. This analysis implies that ensuring enough reactive support from upstream transmission system is paramount importance when smart inverters in reactive modes. 5.4. Number Tap Switching Operations number tap operations increase on implementing smart inverter s. A significant increase in switching operation is observed when inverters are in Volt-VAR mode. increased number tap operations may deteriorate taps banks, incurring an increased maintenance replacement cost. Note that in this framework, both legacy devices inverters are using local settings local information. Coordinating among several devices could be a potential approach to avoid unnecessary switching operations. This, however, will require implementing additional communication interface between substation, banks, s, inverters.

Inventions 2017, 2, 28 18 20 Based on above observations, each voltage approach has its own pros cons. A legacy is easy to implement, requires no additional investments, does not increase reactive dem. However, because delayed-response, it may result in overvoltage for a few minutes. Thus, legacy may be suitable for feeders where short-term primary voltage violations are acceptable. Smart inverter is instantaneous, thus not resulting in any lag in operation. Furrmore, smart inverter helps in better regulation both primary secondary voltages, however, at cost additional reactive dem. refore, utilities with strict requirements for both primary secondary voltage violations may adopt smart inverter. choice between constant factor Volt-VAR will depend upon requirement for secondary voltage regulation available reactive support. If requirements for secondary voltages limits are strict, n utilities may implement Volt-VAR. A constant factor can be used orwise. Finally, a coordinated may be implemented to decrease number switching operations legacy devices. Implementing a coordinated becomes imperative with s in advanced voltage modes. 6. Conclusions increasing rate customer-scale deployment coupled with grid modernization efforts, calls for revising framework for evaluating impacts on feeder voltages. With increased communication capabilities, integration analysis framework should incorporate impacts voltage on feeder s hosting capacity. This paper evaluates several voltage methods, including both feeder s legacy voltage devices smart inverters. A detailed comparison is made by simulating eight cases studies by (1) only ling legacy devices; (2) only ling smart inverters; (3) ling both legacy devices smart inverters. For given test feeder, a large number deployment scenarios are simulated by varying locations, size, customer penetration level. voltage methods are implemented, ir effectiveness in mitigating impacts is quantified using multiple customer system-level metrics based on both snapshot time-series analysis. It is observed that ling eir legacy devices or smart inverters increase feeder s hosting capacity. feeder can accommodate 100% penetration for all voltage cases except for Case 3 (smart inverter only at PF = 0.995), as compared to 40% with no voltage. For Case 3, hosting capacity increases from 9.3 to 17.1 MW. Legacy, although mitigates primary feeder voltage problems, is not as effective in regulating secondary feeder voltages. Smart inverter results in a better regulation secondary customer voltages. Furrmore, due to time-delayed response legacy, a voltage violation may be observed for a small time-interval until activates. smart inverter is instantaneous. However, it comes at cost increased reactive dem an increased number switching operations requiring feeder upgrades additional maintenance cost. reactive dem is lower for factor (1.3 MVAR for PF = 0.995 3.8 MVAR for PF = 0.98) but much higher for Volt-VAR (6.8 MVAR). total number operations over 1-year period increases from 455 operations (without smart inverter ) to 491 at PF = 0.995, 1243 at PF = 0.98, 1991 for Volt-VAR mode. refore, a suitable smart inverter should be identified depending upon additional factors. In order to avoid unnecessary switching operations, a coordinated approach instead local method can be sought. Acknowledgments: author acknowledges contributions Surya Santoso in nature guidance discussions for assessing impacts hosting capacity. Conflicts Interest: authors declare no conflict interest.

Inventions 2017, 2, 28 19 20 Abbreviations following abbreviations are used in this manuscript: C i pen Customer penetration level i% customers equipped with i pen penetration level deployment in kw corresponding to C i pen S Set discrete customer penetration levels indexed by i, S {1, 2,..., i,... 100} x i deployment scenarios jth deployment scenario corresponding to ith customer j penetration level V i max,k Set maximum primary voltages recorded for k deployment scenarios simulated at ith customer penetration level k-run MCS Monte Carlo study with k scenarios simulated at each customer penetration level H 1,k First-hosting capacity based on k-run MCS H 100,k All-hosting capacity based on k-run MCS S Apparent rating smart inverter connected to panel P Active generation panel Q available Available reactive from smart inverter connected to panel References 1. Barbose, G.U.S. Renewables Portfolio Stards 2016 Annual Status Report; Lawrence Berkeley National Laboratory: Berkeley, CA, USA, 2016. 2. U.S. Department Energy (DOE). SunShot Vision Study; U.S. Department Energy (DOE): Washington, DC, USA, 2012. 3. International Energy Agency (IEA). Technology Roadmap: Solar Photovoltaic Energy; International Energy Agency (IEA): Paris, France, 2014. 4. Peças Lopesa, J.A.; Hatziargyrioub, N.; Mutalec, J.; Djapicc, P.; Jenkinsc, N. Integrating distributed generation into electric systems: A review drivers, challenges opportunities. Electr. Power Syst. Res. 2007, 77, 1189 1203. [CrossRef] 5. Katiraei, F.; Aguero, J.R. Solar integration challenges. IEEE Power Energy Mag. 2011, 9, 62 71. [CrossRef] 6. Liu, Y.; Bebic, J.; Kroposki, B.; de Bedout, J.; Ren, W. Distribution system voltage performance analysis for high-penetration. In Proceedings IEEE Energy 2030 Conference, Atlanta, GA, USA, 17 18 November 2008; pp. 1 8. 7. Liu, E.; Bebic, J. Distribution system voltage performance analysis for high-penetration photovoltaics. In GE Global Research Niskayuna NY; NREL/SR-581-42298 Report; U.S. Department Energy (DOE): Washington, DC, USA, 2008. 8. Tonkoski, R.; Turcotte, D.; El-Fouly, T.H.M. Impact high penetration on voltage priles in residential neighborhoods. IEEE Trans. Sustain. Energy 2012, 3, 518 527. [CrossRef] 9. Alam, M.J.E.; Muttaqi, K.M.; Sutanto, D. An approach for online assessment rotop solar impacts on low-voltage distribution Networks. IEEE Trans. Sustain. Energy 2014, 5, 663 672. [CrossRef] 10. Pezeshki, H.; Wolfs, P.J.; Ledwich, G. Impact high penetration on distribution transformer insulation life. IEEE Trans. Power Deliv. 2014, 29, 1212 1220. [CrossRef] 11. Agalgaonkar, Y.P.; Pal, B.C.; Jabr, R.A. Distribution voltage considering impact generation on tap changers autonomous regulators. IEEE Trans. Power Syst. 2014, 29, 182 192. [CrossRef] 12. Toliyat, A.; Kwasinski, A.; Uriarte, F. Effects high penetration levels residential photovoltaic generation: Observations from field data. In Proceedings International Conference on Renewable Energy Research Applications (ICRERA), Nagasaki, Japan, 11 14 November 2012; pp. 1 6. 13. Hoke, A.; Butler, R.; Hambrick, J.; Kroposki, B. Steady-state analysis maximum photovoltaic penetration levels on typical distribution feeders. IEEE Trans. Sustain. Energy 2013, 4, 350 357. [CrossRef] 14. Shayani, R.A.; de Oliviera, M.A.G. Photovoltaic generation penetration limits in radial distribution systems. IEEE Trans. Power Syst. 2011, 26, 1625 1631. [CrossRef] 15. Debruyne, C.; Desmet, J.; Vanalme, J.; Verhelst, B.; Vanalme, G.; Vevelde, L. Maximum injection acceptance in a residential area. In Proceedings International Conference on Renewable Energies Power Quality, Granada, Spain, 23 25 March 2010; pp. 1 6.

Inventions 2017, 2, 28 20 20 16. Navarro, A.; Ochoa, L.F.; Rles, D. Monte Carlo-based assessment impacts on real UK low voltage networks. In Proceedings IEEE Power Energy Society General Meeting, Vancouver, BC, USA, 21 25 July 2013; pp. 1 5. 17. Smith, J. Stochastic Analysis to Determine Feeder Hosting Capacity for Distributed Solar Pv; EPRI Technical Report 1026640; EPRI 3420 Hillview Avenue: Palo Alto, CA, USA, 2012. 18. Broderick, R.J.; Quiroz, J.E.; Reno, M.J.; Ellis, A.; Smith, J.; Dugan, R. Time Series Power Flow Analysis for Distribution Connected Generation; SAND2013-0537; Sia National Laboratories: Livermore, CA, USA, 2013. 19. Dubey, A.; Santoso, S.; Maitra, A. Understing photovoltaic hosting capacity distribution circuits. In Proceedings IEEE Power & Energy Society General Meeting, Denver, CO, USA, 26 30 July 2015; pp. 1 5. 20. Dubey, A.; Santoso, S. On estimation sensitivity analysis distribution circuit s photovoltaic hosting capacity. IEEE Trans. Power Syst. 2016, 32, 2779 2789. [CrossRef] 21. Jothibasu, S.; Santoso, S. Sensitivity analysis photovoltaic hosting capacity distribution circuits. In Proceedings IEEE Power Energy Society General Meeting, Boston, MA, USA, 17 21 July 2016; pp. 1 5. 22. Jothibasu, S.; Santoso, S.; Dubey, A. Determining hosting capacity without incurring grid integration cost. In Proceedings North American Power Symposium (NAPS), Denver, CO, USA, 18 20 September 2016; pp. 1 5. 23. Padullaparti, H.V.; Chirapongsananurak, P.; Hernez, M.E.; Santoso, S. Analytical approach to estimate feeder accommodation limits based on protection criteria. IEEE Access 2016, 4, 4066 4081. [CrossRef] 24. American National Stard Institute. For Electric Power Systems Equipment-Voltage Ratings (60 Hertz); ANSI C84.1-2011; American National Stard Institute: Washington, DC, USA, 2011. 25. Erdman, W. Secondary network distribution systems background issues related to interconnection distributed resources. Natl. Renew. Energy Lab. 2005. [CrossRef] 26. Victor, L.; Kay, M.; Povey, I. Reverse flow capability tap changers. In Proceedings 18th International Conference Exhibition on Electricity Distribution (CIRED), Turin, Italy, 6 9 June 2005. 27. Tushar, M.H.; Assi, C. Volt-VAR through joint optimization bank switching, renewable energy, home appliances. IEEE Trans. Smart Grid 2017, in press. [CrossRef] 28. Ahmadi, H.; Martí, J.R.; Dommel, H.W. A framework for Volt-VAR optimization in distribution systems. IEEE Trans. Smart Grid 2015, 6, 1473 1483. [CrossRef] 29. Turitsyn, K.; Sulc, P.; Backhaus, S.; Chertkov, M. Options for reactive by distributed photovoltaic generators. Proc. IEEE 2011, 6, 1063 1073. [CrossRef] 30. Padullaparti, H.V.; Nguyen, Q.; Santoso, S. Advances in Volt-VAR approaches in utility distribution systems. In Proceedings IEEE Power Energy Society General Meeting, Boston, MA, USA, 17 21 July 2016; pp. 1 5. 31. Seuss, J.; Reno, M.J.; Broderick, R.J.; Grijalva, S. Improving distribution network hosting capacity via smart inverter reactive support. In Proceedings 2015 IEEE Power & Energy Society General Meeting, Denver, CO, USA, 26 30 July 2015; pp. 1 5. 32. Smart Grid Resource Center. Electric Power Research Institute. Available online: http://smartgrid.epri. com/simulationtool.aspx (accessed on 15 February 2017). 33. Dugan, R.C. Reference Guide. Open Distribution System Simulator (OpenDSS); Electric Power Research Institute, Inc.: Palo Alto, CA, USA, 2016. 34. Reno, M.J.; Coogan, K. Grid Integrated Distributed (Grid), Version 2; SAND2014-20141; Sia National Laboratories: Livermore, CA, USA, 2014. 35. CPR, Clean Power Research: Solarany Where. Available online: https://www.solaranywhere.com (accessed on 15 February 2017). 2017 by author. Licensee MDPI, Basel, Switzerl. This article is an open access article distributed under terms conditions Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).