STOCHASTIC ESTIMATION OF FEEDER-SPECIFIC DISTRIBUTED GENERATION (DG) HOSTING CAPACITY

STOCHASTIC ESTIMATION OF FEEDER-SPECIFIC DISTRIBUTED GENERATION (DG) HOSTING CAPACITY Estorque, L.K.L, REE, MSEE, Manila Electric Company (MERALCO), lklestorque@meralco.com.ph ABSTRACT The significant increase of Distributed Generation (DG) in the network will challenge the utility s process to respond with the proposed DG interconnection in a transparent and timely manner and to ensure the safe and reliable operation of the distribution network. For effective DG planning, especially for high penetration scenarios, utilities need to consider proactive solutions in an effort to estimate the maximum allowable capacity of DG without violating the technical standards of the network. In this paper, an alternative method is developed to estimate the feeder DG hosting capacity of the distribution network. This investigates the impact of utility-scale and customer-based generation on feeder technical issues such as voltage regulation, thermal loading, and fault levels, which limit the DG hosting capacity of the distribution network. The application of this method is not intended to replace detailed impact studies, however this would help distribution utilities in screening DG interconnection applications that will be required to undertake detailed impa ct studies. In effect, this will benefit distribution utilities through improvement in screening methods and will employ proactive approach to DG integration through determination of areas with or without sufficient capacity in accommodating DGs. KEYWORDS: Distributed Generation, Hosting Capacity, Stochastic Estimation, LV network, MV network 1. INTRODUCTION Traditionally, an electric power distribution system is planned and designed to operate radially, hence electricity flows in one direction: from high voltage utility source down to the end-users. With the proliferation of distributed generating resources in the network, there is a paradigm shift from a traditional centralized generation model to a distributed and localized power generation scheme. The significant adoption of Distributed Generation (DG), power generation at the point of consumption, has changed several aspects on the utility business model and various technical processes such as system planning, protection coordination, and other operational requirements (Math H. Bollen, 2011). In general, DG provides a wide range of benefits to customers and to distribution utilities through improvement in distribution system performance levels such as reliability and power quality. However, integration of DG at non-optimal conditions could raise a number of technical issues in the operation of the distribution network. Nowadays, distribution utilities have been overwhelmed by the increasing number of interconnection requests of DGs in the network. The increasing development of DGs in the network is primarily attributed to the implementation of support schemes (feed-in tariffs, net metering, etc.) so as to promote DGs, especially those using clean energy sources such as solar and wind (CIGRE, 2014). Under these circumstances, distribution utilities are experiencing strong pressure to respond timely with the proposed DG interconnections, while at the same time ensuring the integrity and reliability of network operation. To address these issues in a timely and effective manner, distribution utilities often applied simplified studies to assess the viability of the proposed DG facilities With these emerging technologies, the evaluation and assessment of proposed DG facilities have become a mandatory procedure for the planning and operation of the distribution network. These methodologies aim to investigate the potential impacts of the DG interconnection and to ensure its safe integration to the network (Lindl & Fox, 2013). Traditionally, distribution utilities often used tiered evaluation procedures, from conservative preliminary screen reviews to detailed and time-consuming technical studies, to assess the impacts of the DG interconnection. However, as DG penetration in the network increases, it is prudent for distribution utilities to have robust tools that will help assess the capability of the network and its requirement. Optimal planning for DG integration has been approached by industry players and researchers through the application of various methods and state-of-the-art techniques (Keane & Ochoa, 2013). Most efforts have led to the assessment of DG hosting capacity, which is defined as the maximum DG penetration or aggregate DG capacity that can be accommodated without causing adverse impact to the network. Hosting Capacity (HC) estimation is, indeed, an optimization problem; wherein, the analysis should return the maximum amount of DG capacity considering the technical issues or constraints in the network such as voltage regulation, thermal loading, short circuit levels, power quality, etc. In this regard, Electric Power Research Institute (EPRI) has led significant research on the area of feeder hosting capacity in an effort to streamline the DG interconnection process (Smith, 2012). Their research is primarily based on the use of stochastic analysis to determine inverter-based PV hosting

capacity on the feeder. Since distribution utilities do not have the complete control for the allocation of these DGs in the network, HC evaluation methods should take into account the uncertainty introduced by the unknown position of potential generators to produce realistic results. This paper is based from EPRI s technical study (Smith, 2012) that is geared towards the development of an alternative method to estimate DG hosting capacity in the feeder. This method is not intended to replace detailed impact studies, however this would help distribution utilities in screening DG interconnection applications that will be required to undertake detailed impact studies. This will also leverage existing DG interconnection processes by providing a fast-track interconnection approach without compromising the accuracy of the results. In effect, this will benefit distribution utilities through improvement in screening methods and will support the development of a proactive solution to DG integration by identifying areas with or without sufficient capacity in accommodating DGs. Moreover, distribution utilities could also redirect the location of customer applications to realize the locational benefit of widespread DG adoption, such as the reduction of technical system losses. In the following sections, the framework used for the evaluation of the feeder DG hosting capacity have been detailed. 2. METHODOLOGY The methodology framework for the development of a tool that will determine feeder-specific DG hosting capacity of the distribution network is shown in Figure 1. The modified IEEE-123 node test feeder is adopted. The modeling and analysis are performed entirely using EPRI s OpenDSS (Open-source Distribution System Simulator). Using the baseline network model, power flow simulations and short circuit analysis are performed for different DG penetration levels. OpenDSS, with Matlab COM interface, is used to perform two DG deployment scenarios: (a) small-scale customer-sited DG at the low voltage (LV) network and (b) utility-scale DG at medium voltage (MV) network. Using stochastic search algorithm, different deployment scenarios are developed. The deployment scenarios consider the uncertainty on the size and the location of the DGs. The performance of the network, for every DG deployment scenario, is recorded and monitored. Eventually, impact assessment is used to analyze the multiple number of simulation results. The results of the analysis determine the feeder-specific DG hosting capacity of the distribution network under a specific condition. Figure 1. Methodology framework To estimate the DG hosting capacity of the network, a co-simulation framework has been developed as shown in Figure 2. The distribution network, loads, and DG technologies have been modeled using OpenDSS. OpenDSS is an open source electric power distribution system simulator which is capable of supporting DG integration and other grid modernization efforts (Dugan, 2013). For simulation, a search algorithm has been developed using Matlab. At this point, OpenDSS has been interfaced with Matlab through the in -process Component Object Model or its COM interface server. At every iteration of the simulation process, technical analyses (power flow and short circuit studies) are performed to calculate voltage profile, thermal loading, and short circuit levels in the network using OpenDSS. Any other distribution system software that can perform unbalanced multi-phase distribution system analysis and has data exchange interface, especially with Matlab, can also be used for solving this problem. The procedure of the implemented method is discussed below while the detailed analysis framework is discussed in the next section. Step 1: Model the base case network using OpenDSS. Step 2: Input the type of simulation. Step 3: Run power flow and short circuit analysis, using OpenDSS, for the base case scenario.

Step 4: Perform DG deployment scenarios based on the input type of simulation using Matlab. Step 5: Run power flow and short circuit analysis for every DG deployment scenario. Step 6: Compute the value of the monitored criteria for every DG deployment scenario. Step 7: The performance of the network for every scenario is monitored and recorded using a spreadsheet file. Step 8: Evaluate the performance results of the DG deployment scenario and determine the Hosting Capacity levels for a given condition. Figure 2. Block diagram of the OpenDSS-Matlab co-simulation framework 3. FEEDER MODELING AND ANALYSIS 3.1 Feeder Modeling The modified IEEE-123 node test feeder (4.16 kv) was adopted and has been modeled using OpenDSS. The feeder model used for the analysis was modified to create a more detailed representation of the actual system. The test system considers an integrated medium voltage - low voltage (MV-LV) network modeling of the distribution network. For that reason, the single-phase MV spot loads were converted to the LV network through the use of service transformers. Due to the significant challenge of accurately modeling the LV network within the MV network, the LV side of the service transformers were used as proxies of the LV cust omers. Since there were no load customers served by the distribution transformer XFM-1 (4.16D/0.48D), this transformer was deemed to be removed in the network. Furthermore, voltage regulators installed along the feeder were removed to simplify the network operation especially during DG deployment scenarios. The topology and the characteristics of the power delivery and conversion elements were partially considered with some exceptions as discussed in Table I. With the help of the GridPV plotting functions (Reno & Coogan, 2013), the feeder topology of the test system used is illustrated in Figure 3. The feeder operates at a nominal voltage of 4.16 kv and the most comprehensive feeder among other IEEE radial distribution test feeders (Kersting, 2001). The substation transformer power rating is 5 MVA which serves 123 primary buses (65 of which are three-phase buses) and 80 single-phase 2.4/0.24 kv service transformers. For voltage regulation and power quality improvement, the feeder

has installed a voltage regulator at the substation and four capacitors in the network. The full load demand of the substation is 3.7 MVA at 0.94 lagging power factor. Three-phase loads are connected in wye or delta configurations while single-phase loads are connected line-to-ground. Moreover, feeder loads are also modeled as constant kw and kvar (PQ), constant impedance (Z), or constant current (I). Table I. Details of the IEEE-123 node test feeder Characteristics Original Model Modified Model OH and UG line segments Fully considered Spot Loads 3Ph and 1Ph at MV network 3Ph at MV; 1Ph at LV Transformers Substation and XFM-1 Removed XFM-1 & deployed service transformers Voltage regulators Four (4) step-type VR One (1) step-type VR at the substation bus Switches Six (6) closed SW and Five (5) Five (5) closed SW and no open opened SW switches Capacitor banks Fully considered Figure 3. Network topology of the modified IEEE-123 node test feeder 3.2 Base Case Analysis The hosting capacity analysis was executed using steady-state analysis. The analysis examined the large variation of DG deployment scenarios and determined the maximum allowable feeder response that would occur with the increasing penetration of DG in the network. This was solved using power flow and short circuit simulations. The power flow study was used to determine the bus voltage profile and loading of the components in the network for different scenarios. On the other hand, the short circuit study was used to determine the impact

of the increasing penetration of DG on the short circuit levels in the network that may affect system protection in real case networks. Other analysis, such as harmonics, stability studies, etc., were not involved in the scope of this study since voltage, loading and short circuit levels are already good indicators of DG impact on the network (Smith, Rylander, Rogers, & Dugan, 2015). Moreover, these analyses were primarily used by distribution utilities in determining possible impacts of DG interconnection. Using the base case scenario, the tap setting of the voltage regulator at the substation bus was determined by performing power flow study. The voltage profile of the base case network without and with voltage regulator action at the substation are shown in Figures 4 (a) and (b), respectively. These figures clearly illustrate the function of the voltage regulator at the substation. Without voltage regulator installed at the substation, undervoltage occurs in the network wherein most of phase voltages, particularly Phase A, drop below the lower statutory limit (0.95 p.u.). As a means to boost voltage or to regulate the voltage at the substation or at any part of the network, the installation of the voltage regulator is required. In the base case network and the power flow study, tap position #6 is the setting for the voltage regulator with the following specifications: Voltage Level = 120; Bandwidth = 2 Volts; PT ratio = 20; Primary CT rating = 700; R-setting = 3; X-setting = 7.5. Furthermore, the performance measurements of the monitored criteria is shown in Table II. This clearly shows that the installation of the voltage regulator, with tap position #6, provides an improvement of the voltage profile, line loading, and aggregate system loss in the network. Figure 4. Voltage profile of the base case scenario (a) without voltage regulator (b) with voltage regulator (Phase A-black; Phase B-red; Phase C-blue) Table II. Network response for the base case scenario Performance Criteria Voltage Regulator (Tap = 0) Voltage Regulator (Tap = 6) Maximum Primary Phase Vpu 1.0000 1.0375 Minimum Primary Phase Vpu 0.9279 0.9659 Maximum Secondary Phase Vpu 0.9969 1.0346 Minimum Secondary Phase Vpu 0.9247 0.9626 Maximum Line Loading 95.04% 93.76% Total Active Power (in kw) 3388.48 3481.88 Total Reactive Power (in kvar) 1271.91 1264.77 Substation Loading (%) 72.39% 74.09% System Loss (%) 2.98% 2.84% Three-phase fault at s/s bus 8,388 A Single line-to-ground fault at s/s bus 8,450 A Line-to-line fault at s/s bus 7,264 A

3.3. DG Deployment Analysis Framework The DG deployment analysis performed is broken down into two categories: the small-scale DG deployment (for customer-based generation) and the large-scale DG deployment (for utility-scale installations). Small-scale DG deployment is situated on individual customers at the low voltage network, while the large-scale DG deployment is based on DG systems interconnecting at the medium voltage network through a step-up transformer. Small-scale and large-scale DG are stochastically deployed and simulated to determine the feeder response under a specific condition. The stochastic nature of the analysis develops thousands of the potential DG deployments to capture the uncertainty introduced by the unknown position and sizes or capacities of potential DG interconnections. The framework for the development of small-scale and large-scale DG stochastic deployment is illustrated in Figure 5. Figure 5. DG Deployment Routine (Smith, 2012) 3.3.1 Small-scale DG deployment The first step in creating small-scale DG deployment scenarios is to collect all possible locations; in our analysis, this includes all LV network customer load buses. The selection of customer load buses follows a uniform distribution, meaning there is an equal likelihood between load buses to be selected. A draw from the pool of locations determined the location of the first DG deployed in the network. Once the load bus is selected, it would be removed from the random pool of additional DG locations in the particular DG deployment scenario. The DG sizes per customer load buses are also determined by a random draw which follows the DG size distribution function, as shown in Figure 6. According to the small-scale DG size distribution, the output DG capacity has a range of 0.5 kw to 100 kw with a mean output capacity of 7 kw. The procedure of simultaneous random locations and DG sizes is repeated until all customer load buses have DG facilities or the maximum allowable aggregate DG capacity has been reached. In our study, there are 80 LV customer load buses with 530 load customers in the modified IEEE-123 node test feeder. The installation of small-scale DGs is reflected by the customer penetration level, defined as the percentage of customers with installed DG systems. Zero percent customer penetration level is the base case scenario means that there is no installed DGs in the system, whereas hundred percent customer penetration level means that all customers have installed DGs. To minimize the simulation time of the analysis, simulation was performed for discrete steps (2%) of the customer penetration levels until the maximum allowable aggregate DG capacity of 5MW has been reached. The aim of the analysis is to generate as many scenarios as possible. Ultimately, five hundred (500) test cases or deployment scenarios were performed. Therefore, with 500 scenarios each with less than 50 simulations, there would be about 25,000 permutations of possible combinations of DG sizes and locations. For each combination, power flow and short circuit simulations were solved to determine the feeder-specific hosting capacity analysis of the network. 3.3.2 Large-scale DG deployment Same simulation approach as the small-scale DG deployment, however this deployment routine used the location of three-phase primary buses as the points of interconnection and at each penetration level, a 100kW DG is interconnected at randomly selected locations with equal likelihood. The 100 kw DG is interconnected through a 415 V three-phase step-up transformer. The DG penetration level is increased until 5MW DG capacity, which is the maximum allowable aggregate DG capacity, is deployed in the network. For a single large-scale deployment scenario, there are 50 penetration levels. Therefore, with 500 scenarios each with 50 simulations, there are 25,000 permutations of possible combinations of DG sizes and locations. For each combination, power flow and short circuit simulations were solved to determine the feeder-specific hosting capacity analysis of the network.

Figure 6. Small-scale DG size distribution 3.4. Feeder response and Hosting Capacity Each feeder response (voltage, loading and short circuit levels) is addressed by determining the respective DG hosting capacity of the network. The hosting capacity is determined when a DG deployment scenario causes the feeder response to exceed the allowable thresholds. In this case, the maximum amount of DG that can be accommodated by the network without violating the technical standards of the network has been met. However, since feeder hosting capacity is based on stochastic analysis, different deployment scenarios are simulated to determine the conservative estimate of hosting capacity given a particular network. The result of a hosting capacity analysis conducted by EPRI (Smith, 2012), which provides insights on how much DG capacity can be accommodated by the network without causing adverse impacts, is shown in Figure 7. According to the results of the hosting capacity analysis, the hosting capacity is defined as a range of the estimated minimum and maximum hosting capacity that determines the more and less conservative estimates, respectively. Region A, values below the minimum hosting capacity, illustrates simulation results that do not cause adverse impacts regardless of the DG locations in the feeder. At the start of Region B, values between the minimum and maximum hosting capacities, the first DG deployment exceeds the monitoring threshold. Moreover, penetration values within Region B are site-specific wherein some simulation results are either acceptable or not acceptable. Region C, values above the maximum hosting capacity, includes all simulation results which exceed the monitoring threshold. In this study, the analysis performed by EPRI is considered to determine its application using the modified test system and other types of DG technologies. However, this study is focused on the determination of the zero risk area () wherein the aggregate DG capacity is less than the minimum hosting capacity; therefore, the analysis returns a conservative estimate of the feeder hosting capacity. Figure 7. Feeder Hosting Capacity (Smith, 2012)

3.5 Monitoring criteria Typical DG planning criteria and limits used in this study have already been employed in international and local practices. The summary of criteria used in the analysis for flagging potential concerns in the DG deployment scenarios is shown in Table III. For the overvoltage criterion, the hosting capacity analysis examined the voltage impact of DG deployment scenarios to the entire network. This includes all buses along the primary and secondary lines. The overvoltage criterion is the primary concern of distribution utility with regards to the integration of DG in the network. The overvoltage caused by DG can be the limiting factor on the amount of aggregate DG capacity that can be safely accommodated in the network. DG can counter the drop caused by the load demand and in high penetration scenarios, this could result in unacceptable high voltages in the network. The limit used for the overvoltage criterion, which is greater than or equal to 105 percent (105%) of the nominal value, is primarily based on local utility practices and studies conducted by EPRI (Smith, 2012). Thermal loading in the network involves the change in the net demand in the feeder with the integration of DG in the network. As DG penetration increases, the net demand decreases and potential reverse power flow conditions may occur (Smith, 2012). Additionally, thermal loading limits are one of the important factors that need to be considered in DG integration. From the power flow analysis, network component (e.g. lines and transformers) ratings were used to determine which DG deployment scenarios have the potential to create thermal loading issues. Three-phase to ground, line-to-line and single-phase to ground faults were examined at the substation bus and at the point of common coupling or PCC. Without DG on a radial system, the fault current is primarily one-directional and flowing from the substation source (Smith, 2012). As regards, highest fault current occurred at the substation bus and decreases as the distance from the substation increases due to the additional system impedance. The limit used for fault current contribution criteria was primarily based on the criteria that the short circuit duty of protective devices should be at least 110% of the maximum fault (ERC, 2011). The 10% conservative limit is designed to ensure that the short circuit duty of the protective devices will not be exceeded with the increasing penetration of DG in the network. However, in real case applications, the utility could actually use the short circuit ratings of the existing protective equipment in the network to determine the allowable limits considering protection issues. But for the purpose of this study and with the absence of detailed planning data, the 10% limit was adopted to simulate the effects of increasing DG penetration on the short circuit levels in the network Table III. Monitoring criteria and limits for the hosting capacity analysis Category Criteria Basis Limit violations Primary Phase Overvoltage Primary Bus Phase 1.05 Vpu Voltage Voltages Secondary Phase Overvoltage Secondary Bus Phase 1.05 Vpu Voltages Max Thermal Loading Line Loading 100% normal Loading rating Max Thermal Loading Transformer Loading 100% normal rating Fault Contribution Total fault current contribution at the 10% increase Protection substation bus and PCC, Reverse Power Flow Reverse Power Flow on the substation transformer Forward Power @ s/s < 0 4. RESULTS AND ANALYSIS The following figures illustrate the feeder response trends from increasing DG penetration on the feeder. These trends are shown for both small-scale and large scale DG deployment scenarios. 4.1 Small-scale DG deployment Overvoltage trends for primary and secondary bus are shown in Figures 8 (a) and (b) respectively. The figures show that the maximum bus phase voltages are seen at the primary buses. Moreover, the figures also identify the minimum hosting capacity which corresponds to the minimum amount of generation based on the occurrence of at least one constraint violation in the network. The minimum hosting capacity bounds the area

which includes all allowable aggregate DG capacity that will not cause any issues to the network. This is termed as the zero risk area which is the allowable DG capacity lower than the minimum hosting capacity (Rossi, Vigano, & Moneta, 2015). For the voltage criteria, the minimum hosting capacity based on primary overvoltage criterion is estimated at 940 kw, whereas the minimum hosting capacity based on secondary overvoltage criterion is estimated at 1,297 kw. Figure 8. Overvoltage Trends (a) Primary Bus (b) Secondary Bus The relationship between the maximum line and transformer loading versus the aggregate small-scale DG capacity are shown in Figures 9 (a) and (b), respectively. It is observed that there were no line loading violations based on the stochastic simulation. The falling edge describes the decrease in line loading since DG has to support local loads in the network, however the rising edge describes the increase in line loading due to the reverse power flow conditions at HV/MV transformer. On the other hand, there are transformer loading violations due to the increased possibility of reverse power flow at the distribution transformers. Increasing level penetration of DGs at the distribution transformer level could cause reverse power flow conditions that could exceed the thermal rating limit of these transformers. For the transformer loading criteria, the minimum hosting capacity is estimated at 1,163 kw.

Figure 9. Loading Trends (a) Line (b) Transformer For the short circuit level violations, the performance of the network is recorded for the singleline-to-ground (SLG) fault current contribution at the substation. SLG fault is chosen since it was observed that in the analysis, SLG faults are greater than three-phase faults due to the installation of single-phase DGs at the LV network. As shown in the Figures 10 (a) and (b), there is a positive and linear relationship between short circuit fault current contribution and the aggregate DG capacity in the network. Furthermore, the large variation in short circuit fault contribution at PCC, as illustrated by Figure 10 (b), means that this criterion is highly site-specific. For the short circuit level criteria, the minimum hosting capacity based on SC contribution at the substation is estimated at 1,844 kw, whereas the minimum hosting capacity based on SC contribution at PCC is estimated at 500 kw. Since SC contribution at PCC is highly dependent on the location of the DG interconnection, this criterion is considered to be a stringent factor for the estimation of hosting capacity.

Figure 10. Fault Level trends (a) at the substation (b) at any PCC 4.2. Large-scale DG deployment Same level of analysis is used for the large-scale DG deployment. Overvoltage trends, loading trends, and fault level trends are determined for the analysis of the large-scale DG deployment scenarios. Furthermore, additional analysis is performed for rotating-machine DG technologies such as synchronous generator and induction generator. This evaluates the effect of different type of DG technologies on the feeder response of the network. This section discusses the feeder response trends for the deployment of large-scale DGs in the network. For inverter-based DG: The overvoltage, loading and fault level trends for the interconnection of inverter-based DGs are shown in Figures 11 (a), (b), (c), and (d). Same as the small-scale DG deployment scenario, the maximum bus phase voltages are seen at the primary buses. Also, it is observed that there were no line loading violations based on the stochastic simulation. However, it is expected that as DG penetration increases, reverse power flow conditions occur; therefore, there will be an increasing probability of line loading violations as DG penetration increases. Moreover, for the large-scale DG deployment, the performance of the network is recorded for the three-phase

fault current contribution at the substation. Three-phase fault is chosen since it was observed that in the analysis, three-phase faults are greater than other type of faults due to the installation of three-phase DGs at the MV network. Figure 11. Large-scale feeder response trends (a) Primary Overvoltage (b) Secondary Overvoltage (c) Line Loading (d) Three-phase fault level at the substation The overvoltage and fault level trends for the interconnection of synchronous generators and induction generators are shown in Figures 12 and 13, respectively. These figures illustrate the impact of interconnecting machine-based DGs in the network. Based on the results of large-scale DG deployment, overvoltage issues are one of the limiting criteria for the interconnection of inverter-based DGs and synchronous generators, whereas it has little impact to the interconnection of induction generators. On the other hand, short circuit level issues, particularly on the substation, are the main concerns for the ins tallation of the rotating machine DGs. Moreover, line loading issues are the least concern for the installation of DGs in the network.

For synchronous generators Figure 12. Large-scale feeder response trends (a) Primary bus phase voltage (b) Three-phase fault level at the substation For induction generators Figure 13. Large-scale feeder response trends (a) Primary bus phase voltage (b) Three-phase fault level at the substation The estimated small-scale and large-scale DG hosting capacities for each of the monitoring criterion are summarized in Tables IV and V, respectively. A default of 5,000 kw minimum hosting capacity is used for the monitoring criterion, since this value is the maximum allowable DG capacity for the feeder. For small-scale DG deployment scenario, unity power factor of inverter-based DGs (solar PV) is used; whereas, for the largescale DG deployment scenario, the following assumptions are used: Inverter-based: unity PF Synchronous generator: 0.9 lag PF, Subtransient reactance (Xd ) of 0.15 Induction generator: 0.9 lead PF, Subtransient reactance (Xd ) of 0.2 Table IV. Small-scale DG Hosting Capacity Category Criteria Limit Minimum HC (kw) Voltage Primary Phase Overvoltage 1.05 Vpu 940 Secondary Phase Overvoltage 1.05 Vpu 1,297 Loading Max Line Loading 100% normal rating (5,000) Maximum Transformer Loading 100% normal rating 1,163 Protection Fault Contribution 10% increase 1,844

Table V. Large-scale DG Hosting Capacity (a) Inverter-based DG Category Criteria Limit Minimum HC (kw) Voltage Primary Phase Overvoltage 1.05 Vpu 1,700 Secondary Phase Overvoltage 1.05 Vpu 4,200 Loading Max Line Loading 100% normal rating (5,000) Maximum Transformer Loading 100% normal rating (5,000) Protection Fault Contribution 10% increase 3,000 (b) Synchronous Generator Category Criteria Limit Minimum HC (kw) Voltage Primary Phase Overvoltage 1.05 Vpu 700 Secondary Phase Overvoltage 1.05 Vpu 2,000 Loading Max Line Loading 100% normal rating (5,000) Maximum Transformer Loading 100% normal rating (5,000) Protection Fault Contribution 10% increase 900 (c) Induction Generator Category Criteria Limit Minimum HC (kw) Voltage Primary Phase Overvoltage 1.05 Vpu (5,000) Secondary Phase Overvoltage 1.05 Vpu (5,000) Loading Max Line Loading 100% normal rating (5,000) Maximum Transformer Loading 100% normal rating (5,000) Protection Fault Contribution 10% increase 1,300 5. CONCLUS IONS AND RECOMMENDATIONS Several methods and approaches have been documented to provide insights on the impact assessment of DG interconnection in the distribution system. Likewise, this work describes a stochastic-based analysis to estimate the DG hosting capacity of the distribution network and to visualize the capability of the network in accommodating proposed DG interconnections. Results show that overvoltage criterion presents a major limitation on the DG hosting capacity of small-scale DG at the LV network. The findings also demonstrate that for small-scale DGs, overvoltage problems will occur before there will be issues concerning short circuit current levels. Whereas, protection issues due to the increased short circuit levels, are more likely to occur in the utility-scale DG deployment, hence limiting the DG hosting capacity of large-scale DG at the MV network. This is mostly true for the installation of rotating-machine DGs such as synchronous generators and induction generators. On the other hand, thermal loading violation is not much of a concern considering that the DG will partly supplied the local load in the system. The results of the feeder-specific hosting capacity analysis could be used as an alternative to conservative screen reviews to determine the true impact of DG integration at the distribution network and to employ the developed methodology in the screening of DG applications. Finally, the application of this method on realistic network models and demand profiles (off-peak and peak loads) is expected to provide better results, as compared with the results obtained using the modified test system. The following are some suggestions for future work related to hosting capacity analysis: One limitation of the study is that the analysis performed are solely based on peak capacity approach. Future research could integrate analysis such as time-series analysis to consider the variation in load and generation profile in the computation of DG hosting capacity. This also emphasized the need for the development of robust tools for the analysis of variable-source DGs. Application of the developed method to real case networks for potential application to actual DG impact studies. The addition of monitoring criteria limits such as harmonic limits (THD, etc.), protection coordination/desensitization, and steady-state and transient voltage variations. However, performing such analysis could entail a lot of resources, in terms of the data needed as well as the validity of the models to be used.

Development of rule-based techniques or procedures based on the results of the hosting capacity analysis to be used as an alternative to traditionally employed and conservative preliminary screen reviews (e.g. 15% peak load, etc.) References CIGRE, (2014). Capacity of Distribution Feeders for Hosting DER. CIGRE. Dugan, R. (2013). Reference Guide: The Open Distribution System Simulator (OpenDSS). Electric Power Research Institute. ERC. (2011). Investor-owned Electric Distribution Utility Planning Manual. Keane, A., & Ochoa, L. e. (2013). State-of-the-Art Techniques and Challenges Ahead for DG Planning and Optimization. IEEE Transactions on Power Systems. Kersting, W. (2001). Radial distribution test feeders. Power Engineering Society Winter Meeting, vol. 2. pp. 908-912. Lindl, T., & Fox, K. (2013). Integrated Distribution Planning: A Proactive Approach for Accommodating High Penetrations of Distributed Generation Resources. Interstate Renewable Energy Council, Inc. Math H. Bollen, F. H. (2011). Integration of Distributed Generation in the Power System 1st Edition. New Jersey: John Wiley & Sons, Inc.,. Reno, M., & Coogan, K. (2013). Grid Integrated Distributed PV (GridPV). Sandia National Laboratories. Rossi, M., Vigano, G., & Moneta, D. (2015). Hosting Capacity of Distribution Networks: Evaluation of the Network Congestion Risk due to Distributed Generation. International Conference on Clean Electrical Power. Smith, J. (2012). Stochastic Analysis to Determine Feeder Hosting Capacity for Distributed Solar PV. California: Electric Power Research Institute, Inc. Smith, J. (2015). Alternatives to the 15% Rule: Modeling and Hosting Capacity Analysis of 16 Feeders. Electric Power Research Institute. Smith, J., Rylander, M., Rogers, L., & Dugan, R. (2015). It's all in the plans: Maximizing the benefits and minimizing the impacts of DERs in an integrated grid. IEEE Power & Energy.