Delft University of Technology. Efficient and Economical integration of EV and PV J. A. Obulampalli

Transcription

1 Delft University of Technology Efficient and Economical integration of EV and PV J. A. Obulampalli

2 ii

3 Efficient and Economical integration of EV and PV By J. A. Obulampalli in partial fulfilment of the requirements for the degree of Master of Science in Electrical Power Engineering at the Delft University of Technology, to be defended publicly on 17 th July 2015 Supervisor: Prof. dr. P. Bauer, TU Delft Dr. ir. Jos van der Burgt, DNV GL Thesis committee: Prof. dr. P. Bauer, TU Delft Dr. L. Ramirez, TU Delft Dr. J. Rueda Torres, TU Delft Dr. Jos van der Burgt, DNV GL iii

4 iv

5 Preface This work was carried out in association with the new-energy technologies group of DNV GL, a consultancy company with its energy division headquartered in Arnhem, Netherlands and TU Delft. DNV GL Energy, which is the Energy vertical of DNV GL (i.e. one of its four Business Areas), has 2300 energy experts and provides services and solutions in the form of Energy Advisory, Laboratories for type testing, Renewables certifications and Research and innovation. DNV GL possesses specialized knowledge in the area of both existing and new energy infrastructures, design and engineering, substation automation systems, connections, stations and of managing and maintaining complex grids. The vision of DNV GL is to make the world safer, smarter and greener for the future. The thesis is a part of "Novel E-MObility Model - NEMO" project, which is funded by the governments of The Netherlands, Germany and Denmark. The idea behind the project is to create integrated software to assess the impact of electrical vehicles on the grid from a technical and economic point of view. The thesis aims at investigating the extent of the asset loading and voltage problems due to integration electric vehicles and photovoltaic into the distribution system and finding the most efficient and economical solution. The thesis mainly focusses to answer the following questions: What is the extent of over-loading and voltage problems due to EV/PV integration in distribution systems? What is the most economical and effective solution for integration of EV/PV among the solutions mentioned in literature? J. A. Obulampalli Delft, July 2015 v

6 Acknowledgement I would like to thank DNV GL and TU Delft for the opportunity they have given me to realize this M.Sc. Thesis at the New Energy Technologies section of DNV GL. During the thesis period I received immense technical guidance and support from Dr. ir. Jos van der Brugt, Thank you for your support and patience during this process. I wish to express my sincere thanks to Dr. Martijn Huibers and Santiago Penate Vera for their guidance and suggestions during the process. I would like to thank Prof. Pavol Bauer for his guidance and for the opportunity to learn and work in his association. I would like to whole heartedly thank, Ir. Gautham Ram for his constant support and suggestions during master thesis. At last I would like to thank and express my gratitude to my parents, my brother and friends for their support and inspiration. -Jagannath Obulampalli vi

7 Contents 1 Introduction Literature review on Integration of EV/PV into the distribution grid EV Integration PV Integration EV/PV Integration Research gap Solutions for integration of EV/PV into the distribution grids Research Questions: Research Methodology Structure of the report Voltage Variation and Mitigation Solutions Network Reinforcement On load tap changer Static VAr control Storage Curtailment of power at point of common coupling (PCC) Reactive power control by PV inverter Q(U), Q(P) Demand response by local price signals SCADA + direct load control Wide area control Prioritisation of Solutions for the integration of PV/EV into the Distribution system Existing grid and Network Reinforcement Description of Case study EV penetration Limits Bogfinkevej: Rindum Mølleby: Conclusion PV Integration limits Bogfinkevej: Rindum Mølleby : Grid Reinforcement vii

8 3.4.1 Cable upgrade: Transformer upgrade: Conclusion Reactive power control Reactive Power and its effects Impact of Cable impedance on reactive power control PV inverters operating in VAr mode PV Inverter Topology Control of Active and reactive power of Inverter Impact of reactive power on EV charging Bogfinkevej distribution grid Rindum Mølleby Distribution grid EV integration limits with reactive power flow Rindum Mølleby distribution grid with upgraded Cables Yearly simulation of Distribution system with high PV and high EV penetration Bogfinkevej Distribution grid Rindum Mølleby Distribution grid Disscussion Conclusion Storage PLATOS Profile simulation and results analysis Optimal Storage location Optimum power and Energy sizing Optimal storage dispatch Simulation Logic Bogfinkevej Distribution system Rindum Mølleby distribution grid Conclusion Conclusion Annex Annex A : Details of network reinforcement for 100% EV penetration in bogfinkevej distribution system Bibliography viii

9 List of tables Table 1:Details of Distribution grids Table 2:EV intergration limit for Bogfinkevej feeder Table 3: EV integration limits for Bogfinkevej Table 4: EV integration limits: Rindum Mølleby Table 5: PV integration limits of Bogfinkevej Table 6: PV Integration Limits Rindum Mølleby Table 7:network upgrade details: bogfinkevej distribution system Table 8: Network reinforcement details Rindum Mølleby distribution system Table 9: Cable types with their R/X ratios Table 10: Bogfinkevj grid data... Error! Bookmark not defined. Table 11: EV integration limits with reactve power control Table 12: Comparison of reactive power control and traditional grid upgrade Table 13: rindum grid summary Table 14: EV integration limits with reactive power flow Table 15: EV integration limits with cable upgraded Table 16: Comparison of reactive power control with traditional grid upgrade Table 17: Rindum EV integration limits with reconfigured transformer position Table 18: Energy Exchange between the DSO and the consumer Table 19: Losses comparision with different Solutions Table 20:Energy Exchange between the DSO and the consumer Table 21: Losses comparision with different Solutions Table 22: Storage details for bogfinkevej diatribution system Table 23: Comparision of solutions for Bogfinkevej Distribution system Table 24: Storage details Rindum Mølleby distribution system Table 25: Comparision of solutions for Rindum Mølleby distribution system ix

10 List of Figures Figure 1:Cable details of the bogfinkevej distribution network before upgrade Figure 2:Cable details of Bogfinkevej distribution system after upgrade for 60% EV penetration Figure 3:Cable details of Bogfinkevej distribution system after upgrade for 100% EV penetration: Figure 4:Cable details of Rindum Mølleby without cable upgrade Figure 5:Cable Details of Rindum Mølleby with cable upgrade for 40% EV penetration Figure 6: EV penetration forecast Figure 7:Circuit with a resistance and reactance to compare the sending in voltage and recieving end voltage Figure 8: Voltage variation with change in power factor Figure 9: Circuit for checkingvoltage variation of different cable types with p.f Figure 10: Graph showing the variation of Voltage with different cable types Figure 11: Schematic of inverter operating in VAr mode Figure 12: Grid connected inverter circuit Figure 13: Equivalent circuit and phasor diagram of single phase inverter Figure 14: Flow chart of PLATOS Figure 15: location of Storage in Bogfinkevej grid Figure 16: location of storage in Rindum Mølleby grid Figure 17:Voltage profile before cable upgrade for feeder 1 with 32 EV s Figure 18:Voltage profile after cable upgrade for feeder 1 with 32 EV s Figure 19: Voltage profile before cable upgrade for feeder 2 with 51 EV s Figure 20:Voltage profile after cable upgrade for feeder 2 with 51 EV s Figure 21: Voltage profile before cable upgrade for feeder 4 with 26 EV s Figure 22:Voltage profile after cable upgrade for feeder 4 with 26 EV s x

11 0

12 1 1 Introduction Electric vehicle (EV) is an element of the future transition towards a clean and sustainable energy system. The sustainability and the impact of electrical vehicles on the environment depends upon the source of electrical energy. Electrical energy produced from fossil fuels has the largest carbon footprint and the energy from renewable sources like hydro, wind and photovoltaic have near zero Green House gases emission during operation. Thus to maximize the positive impact of electrical vehicles on the environment, electrical energy should be sourced efficiently from renewable sources. The most efficient way to deliver energy is to generate energy at the load center to avoid transmission losses, and Distribution generation does exactly the same. The existing electrical grid is designed for Centralized generation and distributed loads and with the integration of renewables, the power flow is altered. When a significant magnitude of distributed power flows from the LV system to the MV system, it can lead to issues like over-voltage, overloading of assets in the transmission system, harmonics, and intermittency in available power. EV s with their ability of decentralized storage of electrical energy, can contribute towards solving the challenge of integrating distributed generation (DG) into existing grids. But their initial deployment into existing grids can cause local problems like under-voltage and peak loading[1] due to tendency of an EV s user to charge during the evening when the household demand is also high The study aims at investigating the extent of the above said problems due to integration Electric Vehicles and Photovoltaics into the distribution system and finding the most efficient and economical solution. Part of the simulation is done using the NEMO tool suite which generates profile, simulates the load flow and optimizes the storage and grid parameters for a complete year. The load flow and grid analysis is performed through DIgSILENT PowerFactory via the PLATOS interface. The Thesis assigned was a part of "Novel E-MObility Model - NEMO" project. The idea behind the project is to create integrated software to simulate and assess the impact of Electrical vehicles on the grid from a Technical and Economic point of view. The software developed is through integration of existing simulation software from the project partners DNV GL - Energy, Fraunhofer ISE and EMD. EnergyPRO from EMD generates the profiles for households, distributed generators and SimTOOL from Fraunhofer ISE performs load flow studies and Demand side management. PLATOS from DNV GL is an optimization tool for determining location, type and size of storage systems in order to avoid voltage or grid overload problems which result from variable distributed generators or electrical loads. 1

13 1.1 Literature review on Integration of EV/PV into the distribution grid Photovoltaic systems convert solar power into DC power and is then converted into grid compatible AC power through the use of an inverter. A PV system can be modelled as a PV node or PQ node for load flow analysis based on the control system of the inverter[2]. If the converter controls Active power (P) and Voltage (V) independently then it s modelled as a PV node and If Active power (P) and Reactive power (Q) are controlled independently then it is represented as a PQ node. Electrical vehicle charger is modelled as a PQ node as it is considered a load. Thus both PV and EV can be considered as a load with different sign convention, EV can be considered a positive load and PV can be considered as a negative load. The voltage across two terminals of a cable is governed by the equation: V = (I active + ji reactive )(R + jx) It can be seen that when EV and PV are modelled as PQ nodes the voltage change depends on the magnitude of active and reactive current flowing through the feeder. Therefore there is a possibility of over voltage (>1.05p.u.) in case of PV generation and under voltage (<0.95p.u.) in case of EV charging. Due to the increased flow of active and reactive currents, the cable and transformer may be overloaded (>100% rated kva).the IEC defines that the supply voltage of 220/400V can be between 0.9p.u. to 1.1p.u. In the simulation below, the voltage limits are considered to be +/-0.05p.u EV Integration The impact of Electrical vehicles on the transmission system is studied by S.W.Hadley [3] based on vehicle characteristics, charging characteristics, time of plug in. The reports concludes that changes to the load profiles due to EV charging and broadly states that the distribution system should be planned and reinforced to accommodate the non-intermittent nature of Electrical vehicle charging. Study of a LV distribution grid in suburban Dublin, Ireland by Richardson[4], compares the impact of controlled charging and uncontrolled charging on the distribution network. In controlled charging the voltage drop across the feeder is evident and same is curtailed by using controlled charging. The controlled charging is optimized based on the voltage limits and thermal limits of the system. In the simulation presented in the paper, voltage limitation is predominant over transformer/cable thermal loading and concludes that the same may not be true for distribution systems with lower thermal capacities and higher charging powers. Evaluation done by Taylor[5], in his paper includes the impact of EV s on the life of the transformer. The life of the transformer is reduced due to the increased loading and increased stress on the dielectric. Voltage unbalance in the distribution system due to single phase charging of electrical vehicles is studied by Farhad[6], which suggests that EV s will have a larger impact on the voltage unbalance for the consumers at the end of the feeder than the consumers near the transformer. In conclusion it can be said that the impact of Electric vehicles on the distribution network is well studied PV Integration The impact of photovoltaic systems on the Distribution network is almost the same as EV due to the fact that they can be considered as PQ nodes with opposite flow of power. One of the differentiating 2

14 characteristics of PV system is its output intermittency due to cloud shading which leads to sudden voltage dip/rise[7]. Baran et al.[8] considers a high penetration of residential PV systems and investigates the consequences of PV on Protection and Voltage regulation. His study concludes that the impact on protection is limited due to fact that inverters can limit their currents during faults and can quickly disconnect from the system. Thomsom[9] through a study of a LV network in Leicester, UK concludes that PV integration limits are constrained by the existing network parameters and that the impact of self-consumption on the transformer loading is limited as the residential peak doesn t coincide with the PV peak. A comparison of urban and rural distribution network by Tonkoski [10], suggests that the voltage rise due to PV would be greater in rural networks due to longer distance between two adjacent houses. PV inverters and EV charging stations are electronic non-linear loads/generators which generate harmonics. Schalabbach[11] s study on PV inverters and Lu Yanxia[12] s study on Electric vehicle charger reiterates the same. Harmonics generated can be removed through active and passive filters and are not major hindrances to EV and PV integration EV/PV Integration It can be observed that problems associated with PV and EV integration into the distribution system are well studied and can be broadly classified into voltage, asset loading, phase voltage unbalance and harmonics. These problems can be addressed individually or through smart operation of EV in combination with PV. The issue of voltage and loading can be addressed by traditional method like network reinforcement in which the assets leading to voltage drop and overloading can either be replaced or reinforced by adding parallel transformer or cables. Another solution studied by Ram[13][14] for voltage support is the use of power electronic assisted On-Load tap changing transformer. OLTC can provide voltage support both in the case of EV charging and PV generation as the tap change can either be positive or negative and its integration with power electronics can provide fast response to PV intermittency. The major drawback is that it doesn t solve the problem of the loading of the assets. Network reconfiguration in distribution systems studied by Baran[15], can be implemented by the use of sectionalizing switches to alter the power flow in the distribution system to decrease the losses and loading of the distribution system. This is usually implemented in MV networks where switching alternatives are present and usually not feasible due to the radial nature of distribution networks. Active power curtailment for PV inverters to avoid over voltage is studied by Tonkoski [16], in which the active power is curtailed as a function of power generated. The study concludes that the inverters downstream on the feeder will have to curtail larger magnitudes of power which may affect their revenue. This solution can be compared to limiting the charging power of Electric vehicles studies by Quian[17] Reactive power control to support voltage in a distribution system can be done through PV inverters and or through the use of shunt capacitors, shunt reactors, synchronous condensers, Static VAr compensators and dynamic voltage restorers. The switching of shunt capacitors and reactors need to be controlled as we need to consume reactive power during PV generation and generate reactive power during EV charging. To avoid the complex control system and installation cost of the capacitors and reactors, we consider PV inverter to be the source/sink of continuous variable reactive power. The inverter reactive power control technologies for voltage support are reviewed by Demirok [18] and a 3

15 new reactive power control method is proposed which is a combination of cosᵠ(p) and Q(U). The proposed reactive control methods gives a higher PV integration limits. Based on the above study, the reactive power control method can be replicated for EV charging as well, to get a higher EV integrati on limit like the study by Huang[16]. Reactive power compensation has its limitations for voltage support in the distribution systems of EU countries as the R/X ratios of under-ground cables in LV system are between 5 to 2[19] and needs significant amount of reactive power with the risk of overloading the cables. Hashemi[20] projects integration of Storage as a solution for voltage quality issues and peak shaving in systems with high penetration of PV and EV. The storage absorbs and stores PV energy and discharges during EV charging. By this charging and discharging process, the voltage problems and loading problems are solved. The simulations show that the energy storage capacity needed for over voltage prevention is dependent on the location of the customer on the feeder. The storage capacity is less for consumers near the transformer. Traube[21] in his paper mitigates PV intermittency through the use of EV charging and discharging, in this process EV is being used as a storage system. This is can be an interesting option for work place charging where the EV s can be charged during peak PV output in the afternoon. The main limitation for using EV s for energy storage in residential distribution systems is the low simultaneity factor as most of the EV are not connected to the residential grid in the afternoon when the PV output is at its maximum Research gap The above mentioned studies and solutions either mitigate the problems of voltage and loading by treating them as individual problems or try to mitigate them by smart operation of EV and PV. There is a lack of research work in terms of using the PV inverters to control reactive power during EV charging. As cited above reactive power compensation has been studied individually either from EV point of view, or from PV point of view. A study of the reactive power capabilities of PV inverters by Ellis[22] states that inverters in principle can provide reactive power capability at zero active power but, the functionality is not standard in the industry due to regulations and standardisation. Maknouninjad[23] in his paper discusses steps involved for operating the inverter in VAr mode during the night. The inverter consumes negligible amount of active power from the grid to pre charge the DC bus capacitor and regulate the DC bus voltage within limits. This reactive power capability can provide voltage support during EV charging which is usually during the evening/night when the PV active power generation is not present. This thesis tries to fill this gap by using PV inverters for EV charging and comparing it with other methods to get the most effective and economic solution in terms of impact on voltage and congestion/loading for integration of electric vehicles and PV systems into LV distribution system. 4

16 1.2 Solutions for integration of EV/PV into the distribution grids The solutions for voltage support and loading mentioned in the above literature study can be broadly classified into : Network reinforcement On-load tap changer Network reconfiguration Active power/ charging power curtailment Reactive power control Storage The solutions can be further classified based on the stakeholder implementing it. The report: Prioritization of technical solutions available for the integration of PV into the distribution grid [24] classifies the solution on the basis of implementation by the stake holders. The following solution are studied in brief in the following chapter and Network reinforcement, Distributed storage and reactive power solutions are studied for their effectiveness in later chapters. DSO Solutions o Network reinforcement o On-load Tap Changer (MV/LV transformer) o DSO Storage o Network Reconfiguration Prosumer Solutions o Prosumer Storage o Self-consumption/generation by tariff incentives o Curtailment of power at point of common coupling (PCC) o Active power control o Reactive power control Interactive solutions ( interaction between DSO and Prosumer) o Demand response price signals o SCADA + Direct load Control o Wide area voltage control 1.3 Research Questions: The thesis mainly focusses to answer the following questions: What is the extent of over-loading and voltage problems due to EV/PV integration in distribution systems? What is the most economical and effective solution for integration of EV/PV among the solutions mentioned in literature? 1.4 Research Methodology Based on the following study each of the solution will be studied in brief and the solutions with high priority namely reactive power control, grid reinforcement, storage will be studied to get the best solution in technical terms and economical terms. To get the efficient and economical solution we follow the below methodology: 5

17 Find the EV integration and PV integration limits for voltage and load violations. Increase the EV integration limits with the use of grid reinforcement. Increase the EV/PV integration limits by use of reactive power control. Increase the EV/PV integration limits by use of Storage. Compare the cost associated with the each method with the traditional grid reinforcement technologies. 1.5 Structure of the report The report is structured in the following way: Introduction of the thesis along with the literature study and key solutions Review of the different solutions and identification of the priority solutions Description of the grid studied and the extent to which EV/PV can be integrated into the existing grid along with study of network reinforcement as a solution for EV/PV integration Study of reactive power control as a solution and its impact in the LV distribution system. Storage as a solution and description of PLATOS tool. Conclusion which compares the network reinforcement, reactive power control and storage as solutions for EV/PV integration. 6

18 2 2 Voltage Variation and Mitigation Solutions As mentioned in the section 1.1, Voltage drop across two terminals of a transmission system is governed by the equation: V = (I active + ji reactive )(R + jx) Therefore it can be said that voltage variation mitigation can be done by varying the magnitude of resistance (R), reactance (X), active current( I active )and reactive ccurrent (I reactive ). Active and reactive current are related to each other through power factor. Let us see how the following parameters change with the different solutions mentioned in the previous chapter 2.1 Network Reinforcement This solution is a traditional method of increasing the network potential to accommodate PV and EV charging. In this method, assets operating close to their thermal rating are reinforced by replacing them with higher power capacity cables and transformers. Cables causing the highest magnitude of voltage variation are replaced with cables of lower resistance and R/X ratio. This solution can also include building of new feeder or substation instead of reinforcing the existing infrastructure Reconfiguring the network for altering the power flow in the system to maintain voltage under limits and avoid loading of assets can be done through the operation of sectionalizing switches. This solution is usually used in MV systems where the topology is usually meshed but operated radially. Reconfiguration of the system is usually done to avoid the outages and needs a communication network to be able to intelligently configure the network. Another type of network reconfiguration is closed-loop operation in which each load/generated is fed from two different sources/sinks. This causes the equivalent impedance of the circuit to decrease, thus leading to a better voltage profile. But, in case of a technical fault the impact on the system will be larger due to larger area of impact and fault impedance and will lead to reliability issues. Advantage: The solution and can be used for mitigating both voltage and loading. It is simple from design and execution point of view Can be readily used where there is a possibility for reconfiguration Disadvantage: High cost of Investment, along with disruption of power supply during reinforcement. 7

19 In rural areas where the distance between two households in comparatively larger, reinforcements of larger magnitude than in urban areas is needed to keep the voltage within limits. Lack of future knowledge about possible loads variation and generators may lead to recurring investments. Need intelligent control and communication between the feeders and the Distribution system for control of sectionalizing switches. 2.2 On load tap changer The use of a series transformer or an on-load tap changer to boost the voltage to compensate for voltage sag is not new and the same can be applied for integration of distributed generation by bucking the voltage across a distribution line. The essence of the method is that the supply voltage V s plus the additional series voltage gives the load bus voltage. V l = V s ± V series The series transformer can be combined with power electronics control for sub-cycle response to voltage variation, power factor control[25], thus giving the capability to control reactive power in the system. Advantages: Can be effectively used in cases where the voltage problems are predominant in comparison to line loading Disadvantage: Installation of Series transformer is not simpler in terms of down time or work needed when compared to traditional grid reinforcement. Most of the transformers presently in operation on the distribution network are not fitted with OLTC Impact on the loading of assets is non-existent. 2.3 Static VAr control Static VAr compensators are fast acting reactive power control equipment which control the reactive component of current in the system. In cases of over-voltage the reactive power is absorbed and during under-voltage the reactive power is generated leading to change in voltage. They are usually used in the MV/HV segments of the transmission systems where the reactance is higher than resistance. Advantage: Effective in voltage support Loading of the assets can be minimized when the reactive power compensators are located outside the substations and point of demand. Disadvantage: Expensive when placed only to provide voltage support The line loading is increased due to reactive power flow, and not usually used in distribution systems with under-ground cables as they have high R/X ratios. Switching of capacitors/reactors provides voltage support in steps 8

20 Communication infrastructure is needed between the feeder far end and transformer for effective control. 2.4 Storage Energy Storage to improve power quality in terms of Voltage Depressions and Power interruptions have been in practice for a long time[26] for industrial plants with critical processes. Its usage in the electrical utilities and distribution systems were limited by the cost, centralized generation, maturity of technology and lack of tools to assess the benefits of Storage technology during planning[3]. In recent times due to growing Distributed energy generation and their presence at the end consumer of the distribution network gives Energy Storage a renewed need and push. Storage technologies integrated into the grid can be used for applications like: Instantaneous Applications (Seconds): Mainly rapid spinning reserve, Primary frequency control, ride-through capability. For these applications, the battery should be able to deliver high power in short periods. Short Term applications (minutes): Secondary and tertiary frequency regulation, Smoothing of power output from renewable energy sources, Demand side management by active and reactive power control. For these applications, the battery should be able to deliver high power and medium energy in short periods. Mid-term (minutes to few hours): load balancing, peak shaving, generation in micro grids. For these applications, the battery should be able to deliver high power and high energy in short periods Long term( few days) : Energy Supply in cases where the demand is for few days in a year and is not economic to build transmission infrastructure. Eg. Festival areas with high demand for a few days. In our study we consider using storage for mitigating the problem of loading and voltage through discharge of battery during peak demand and low local generation and charging the battery during peak generation and low consumption. For storage to be effective for voltage, the storage has to be distributed to avoid overloading of lines and voltage drop. Advantages: Can be effectively used for decreasing the loading of assets Voltage support can be realized by either distributing the storage or interfacing them with an inverter to either absorb or generate reactive power Disadvantage: Expensive when compared to other solutions Lack of regulations/standardization regarding the ownership of Storage Needs energy residential energy management systems to optimally operate the battery 2.5 Curtailment of power at point of common coupling (PCC) In this solution, the prosumer is responsible for limiting his consumption or generation based upon the strength of the network (i.e. in case of congestion or voltage problems). This can be done through use of home-based energy storage or by demand response, e.g. charging an electric vehicle at the time of PV 9

21 peak output. Controllable curtailment needs installation of smart meters and residential energy management systems which implies additional investment to the prosumer. Advantages: Effective for both voltage support and loading Disadvantages: Needs regulatory changes as present regulations like the RES Directive 1 in Germany gives unlimited priority to renewable energy Suitable for PV systems as the peaks occur only for a few instances in a year, but not for EV charging, because the charging level is consistent throughout th year and may need additional reinforcement. 2.6 Reactive power control by PV inverter Q(U), Q(P) PV Inverters are able to control active and reactive powers being generated or consumed by them. Therefore it is possible to consume reactive power from the system during PV generation and gene rate reactive power during EV charging. This creates an opposite flow of reactive power which can control the voltage in the system. But this additional flow of power in the system can create over loading of the system as the amount of reactive power required to effect the voltage like active power is almost 2 to 5 times the active power. In this method, the reactive power can be a function of the active power generated, Q(P), or as a function of the voltage of PCC, Q(U). Advantages: Can be an effective solution where Voltage problems are predominant over asset loading like in rural distribution systems. Solution is technically available and doesn t need any additional grid investment Disadvantages: Not Suitable for networks operating near rated thermal loading Present Regulations only recommend reactive power control of systems above 30kWp Inverter reactive control is not a standard in industry 2.7 Demand response by local price signals The effective impact on voltage and loading of assets in the network can be reduced by demand side management (DSM) in which the peak generation or consumption can be controlled by spreading them out in time. The DSM is effected through the use of price signals sent by the DSO or indirectly by TSO based on the network limitations, demand and generation forecasts. In the paper by Shengrong Bu[27] different electricity prices are defined within the DSO network according to grid loading. 1 The RES directive stipulates that : 1. To the extent required by the objectives set out in the Directive, the connection of new RE installations should be allowed as soon as possible. In order to accelerate grid connection procedures, Member states may provide for priority connection capacities for new installations producing electricity from renewable energy sources. (recital (61)). 2. Priority access and guaranteed access for electricity from RES are important for integrating RES into the internal market in electricity. Priority access to the grid provides an assurance given to the connected RES-E generators that they will be able to sell and transmit the RES-E in accordance with connection rules all the times, whenever the sources becomes available. In the event that RES-E is integrated into the spot market, guaranteed access ensures that all the electricity sold and supported abtains access to the grid (rectial(60)). 10

22 This solution requires the installation of smart energy systems which can receive the price information from the system operator, control the residential loads of the consumer based on his settings and give feedback about the system to the DSO. Advantages: Effective in controlling both Voltage problem and loading Its first step towards an intelligent grid Disadvantages: Needs a strong communication network Parameterization of prices for different distribution areas is difficult Prosumers should be protected from being discriminated on the basis of network strength. 2.8 SCADA + direct load control An alternative to demand side management through price signals is direct load control by the DSO which can include controlling the PV generation or by controlling the charging magnitude and time of an electric vehicle. An additional payment can be introduced for customers to get flexibility in load control. This solution needs the installation of smart meters and energy management systems that can be managed by SCADA (supervisory control and data acquisition) which can provide remote access of the loads to the DSO. Advantages: Removes the complexity of assigning prices based on network capacity Can be used for effectively for peak load mitigation Disadvantages: As mentioned in other solutions, a strong communication network is needed which increases the investment on the network Not feasible for fast voltage variations User loses control of their loads and this might not be favorable to them DSO interferes with market parties (customer and supplier), which is not allowed according to the rules of unbundling 2.9 Wide area control This solution is a combination of the above mentioned voltage and reactive power control like Transformer OLTC, distribution voltage regulators and reactive power control through PV inverters. The solutions are coordinated to optimize voltage and power factor for the complete distribution system through voltage and power factor measurements at several points in the distribution system. The components required for this control are available but their integration needs effort both in technical and economical terms. The effect on voltage is evident but its impact on the loading is minimal and in some cases cause overloading due to flow of reactive power. Advantages: Coordinated approach between the different voltage control equipments can lead to better results Disadvantages: Proper coordination between the different devices makes it difficult to implement 11

23 2.10 Prioritisation of Solutions for the integration of PV/EV into the Distribution system. The discussed solutions are evaluated based on the following criteria: Impact on voltage Impact on congestion Technology readiness Solution Network Reinforcement On load tap changer/series Transformer Static VAr control Storage Curtailment of power at PCC Reactive power control by PV inverter Demand response by local price signals SCADA + Direct load control Wide area Voltage control Impact on Voltage High impact (3) High impact (3) Medium impact (2) High impact (3) High impact (3) High impact (3) High impact (3) High impact (3) High impact (3) Impact on asset loading High impact (3) Negligible Impact (1) Negligible Impact (1) High impact (3) High impact (3) Medium Impact (2) High impact (3) High impact (3) Negligible Impact(1) Technical feasibility in Implementation Immediate (3) Immediate (3) Immediate (3) Immediate (3) Near future (2) Immediate (3) Future (1) Future (1) Future (1) Priority High (9) Medium (7) Low (6) High (9) Medium (8) Medium (8) Low (7) Low (7) Low (5) It can be seen that Network reinforcement and storage have the highest priority based on the impact on voltage, loading and ease of implementation. From the medium priority solutions, we select reactive power control through inverter as the next feasible solution as implementation of curtailment of power and Active power control needs changes to the existing directives and is a deterrent to the installation of renewable energy systems. Therefore the solutions that will be discussed in detailed and compared are: Network reinforcement Reactive power control through PV inverters Storage 12

24 3 3 Existing grid and Network Reinforcement The analysis and simulations to find the existing possible integration levels of EV and PV were performed on two LV grids from the Danish Stakeholder of the NEMO project. The grids were modeled in Power factory and simulated for a complete year in hour time steps using varying levels of EV and PV penetration to assess the impact on the power quality. Normal (or slow) domestic charging of Electric Vehicles (EV) is assumed to consume a power of 3.7kW and takes 6 hours to charge 22.2 kwh battery. This is a constant demand unlike other household appliances, and potentially causes under-voltages (<0.95 p.u) due to increased voltage drop in the cables and overloading of transformers due to increased demand. Distributed generation (DG) on the other hand may lead to over-voltages (>1.05p.u.) and transformer overloading due to flow of power from the distribution grid to the transmission grid. The following constraints were considered while simulating the grids: Power Loading limit : 100% of rated value Under-voltage <0.95p.u. Over-voltage >1.05p.u. The IEC defines that the supply voltage of 220/400V can be between 0.9p.u. to 1.01p.u.. For our simulation we consider 0.95 and 1.05 as the limits to account for the voltage variations from the HV and MV side. Table 1:Details of Distribution grids Bogfinkevej Rindum Mølleby Construction 1970 s 2000 s Transformer 250kVA 400kVA Cables (3-phase) 150mm 2 : 0.21km 95mm 2 : 0.51km 50mm 2 : 1.94km 150mm 2 : 2.55km 95mm 2 : 0.55km Distance between the substation and last house connected to the feeder Feeder 1 : 0.41km Feeder 2 : 0.38km Feeder3 : 0.38km Feeder4 : 0.26km Feeder1 : 1.04km Feeder2 : 0.546km Feeder3 : 0.687km Consumption per year 458MWh 399 MWh Nr. of connected houses The two distribution networks are ideal for studying the interaction of EV and DG as they offer different limitations for EV/DG penetration in terms of transformer loading and feeder cable length. Bogfinkevej 13

25 which supplies to 141 households has a relatively less transformer capacity of 250kVA and Rindum Mølleby which supplies to 93 households has a long feeders leading to higher voltage drop. 3.1 Description of Case study It is expected that intelligent operation of EV charging points and Distributed Generation (DG) can solve the above mentioned problems of voltage level and transformer loading while keeping the grid reinforcement investment to a minimum. To cross check the above mentioned solution we follow the following procedure: Check the EV integration limits and DG integration limits of the existing grids Reinforce the grid to accommodate 40%, 60% and 100% EV penetration, to get an idea about grid reinforcement investment. The household load, EV charging load and PV generation profiles for the si mulation are taken from EnergyPRO by EMD International A/S. 3.2 EV penetration Limits We need to get the maximum number of vehicles that can charge from an existing distribution grid to analyse the capacity of the present existing grid. To get the maximum worst case number of EV s that can be charged, we take into account time instance on a winter day ( hrs) when the house hold load is at its peak and EV charging imposes an additional load Bogfinkevej: For feeder 1 which has 32 houses, let us suppose each house may have an EV, there can be 2^32 possibilities in which the EV s can be distributed. To avoid huge computation, we increase the number of EV s from the substation to the end of the feeder till cases with under-voltages arise. The same is done from the feeder far end. Table 2:EV intergration limit for Bogfinkevej feeder 1 Charging Scheme EV placement Max. No. of vehicles charging Charging without delay Near the transformer 16 (charging at 6pm) Far end of the feeder 8 Charging with delay Near the transformer 19 (Charging at 1am) Far end of the feeder 10 By taking the average of the maximum number of vehicles charging from the transformer end and the feeder far end we can get the limiting number of electrical vehicles that can charge from a particular feeder. We can see that charging without delay can accommodate 12 EV s and charging with time delay can accommodate 14 EV s which correspond to 31% and 46% of EV penetration. The increase in EV charging limit from 6pm to 1am is not large as an EV charging power is high as compared to the general house hold demand. 14

26 For other feeders the same method of determining the limit was applied. The EV penetration of feeder 2 is better due to its topology in which many houses are connected near the transformer through branches. The numbers of houses connected to the feeder 2, feeder 3 and feeder 4 are 51, 32 and 26 respectively. Table 3: EV integration limits for Bogfinkevej Feeder Details EV placement Max. No. of vehicles charging Feeder 1: 32 Houses Near the transformer 16 Far end of the feeder 8 Feeder 2 : 51 Houses Near the transformer 41 Far end of the feeder 13 Feeder 3: 32 houses Near the transformer 19 Far end of the feeder 15 Feeder 4 : 26 Houses Near the transformer 17 Far end of the feeder 24 All the feeders combined Near the transformer 88 Far end of the feeder 60 When transformer overloading is not considered, the number of vehicles that can be charged from the complete distribution system without voltage violations under worst case scenario of EV positioning (ie. Charging from the feeder far end) is 60, which accounts to 42% of EV penetration. When transformer overloading is considered the maximum number of vehicles that can be charged from the complete distribution system is restricted to 37 EV s or 26.2% of EV penetration Rindum Mølleby: The distribution system of Rindum Mølleby looks to be strong when compared to Bonfinkeveg as its transformer and cables have a higher rating and the number of houses it supplies to is less. This is not actually the case, Bogfinkevej has an ability of accommodate a higher penetration of EV than Rindum Mølleby due to the fact that the length of all the feeders in Bogfinkevej are almost half that of Rindum Mølleby Table 4: EV integration limits: Rindum Mølleby Feeder Details EV placement Max. No. of vehicles charging Feeder 1 Near the transformer 1 34 houses Far end of the feeder 0 Feeder 2 Near the transformer houses Far end of the feeder 12 Feeder 3 Near the transformer houses Far end of the feeder 12 All feeders combined Near the transformer 25 Far end of the feeder 24 15

27 Rindum Mølleby has a combined EV penetration limit of 18.2% limited by the voltage drop across the feeder. If the under-voltage due to voltage drop in cables is not considered then the EV penetration limit is limited by the transformer loading to 82 EV s or a penetration level of 89% Conclusion It can be said that in case of Bogfinkevej distribution grid, the main limiting factor for EV Penetration is the transformer loading and in case of Rindum Mølleby the limiting factor is the cable voltage drop due to long length of the feeders. 3.3 PV Integration limits PV/DG integration limit is defined as the maximum power that the DG can generate without violating the voltage levels, line thermal capacity and substation transformer capacity. The PV integration limit is defined by considering the worst case scenario when the DG generation is at its maximum and the load is at its minimum. It is assumed that the aim of a domestic PV system owner is to be able to produce enough energy for charging EV. The worst case EV demand is 6000 kwh/year. Denmark experiences yearly average of 3.15 sun hours per day (The Nordic Folkecenter for Renewable Energy, 2014) 4, So the average PV sizing should be 5kWp. PV System peak output = Yearly consumption(kwh) [365 Average sun hours] Bogfinkevej: To get the PV integration limits we start placing 5kW PV Systems from the transformer side to the feeder end and vice versa. The time instant simulated is on a summer day ( ) at 12pm when the PV system produces maximum power and the household load is at its minimum. The PV systems are placed until voltage violations are observed. Table 5: PV integration limits of Bogfinkevej Feeder Details PV positioning PV penetration in kw Feeder 1 Near the transformer 95 Far end of the feeder 75 Feeder 2 Near the transformer 185 Far end of the feeder 85 Feeder 3 Near the transformer 95 Far end of the feeder 80 Feeder 4 Near the transformer 105 Far end of the feeder 70 The worst case PV integration limits for the complete distribution system system is 310kW but the transformer capacity restricts the capacity to 250kW Rindum Mølleby : 16

28 The same method of estimating the PV integration limits used in bogfinkevej is implemented in for Rindum Mølleby and the results are presented in the table below. Table 6: PV Integration Limits Rindum Mølleby Feeder Details PV positioning PV penetration in kw Feeder 1 Near the transformer 60 Far end of the feeder 40 Feeder 2 Near the transformer 80 Far end of the feeder 75 Feeder 3 Near the transformer 100 Far end of the feeder 60 In worst case scenario the maximum PV generation that can be integrated without any over-voltages is 175kW and is restricted by the cable voltage drop. 3.4 Grid Reinforcement The idea behind upgrading the grid was to analyze the magnitude of grid reinforcement required to integrate Electric Vehicles and DG without dispatchable loads, storage, reactive power control etc. The grid was upgraded for EV penetration levels of 40%, 60% and 100% Cable upgrade: The aim of the cable upgrade is to get an estimate on the cable length that needs to be upgraded to be able to accommodate EV charging at 40%, 60% and 100%. Transformer over-loading is not considered as the limiting factor while running the simulations. The EV charging load is superimposed on the peak house hold demand. To simulate the distribution grid response, we select a time instant when the household demand is at its peak on a winter day ( :00). The EV charging load is imposed on the household demand and the voltage profile is plotted. Based on the voltage drop observed, the cables with relatively larger voltage drop is upgraded and again simulated until the voltage drop is above 0.95 p.u Bogfinkevej For 40% EV penetration level, no cable upgrade was needed. For 60 % penetration, 340m from the total cable length of 2.72km has to be upgraded which is 12% of the total cable under use. 950m of cable was upgraded from a total cable length of 2.72km to keep the voltage within limits for a penetration level of 100%. This can be a possible solution where the feeder is strong as in case of feeder 4 where only 26.4m of cable had to be changed to accommodate 100% EV penetration. The details of the cable upgrade for Bogfinkevej distribution system with 100% EV integration are in annexure A. 17

29 Table 7:network upgrade details: bogfinkevej distribution system Traditional grid upgrade 40% EV penetration Transformer upgrade : 400kVA 60% EV penetration Transformer upgrade: 500kVA Cable upgrade: 340m from 2.72km 90% Ev penetration Transformer upgrade:600kva Cable upgrade: 950m from 2.72km Figure 1:Cable details of the bogfinkevej distribution network before upgrade 18

30 Figure 2:Cable details of Bogfinkevej distribution system after upgrade for 60% EV penetration 19

31 Figure 3:Cable details of Bogfinkevej distribution system after upgrade for 100% EV penetration: Rindum Mølleby: For 40% EV penetration, a cable upgrade of 1.7km is needed from a total cable length of 2.81km. For 60% EV penetration, all the cables must be upgraded to 400mm 2 from their existing cable ratings and the grid needs to be reconfigured due to the large lengths of the feeder. Simulation for 100% EV penetration was not considered due to the length of the feeders and need for complete reconfiguration of the grid. 20

32 Table 8: Network reinforcement details Rindum Mølleby distribution system Traditional grid upgrade 40% EV Penetration Cable upgrade : 1.7km from 2.9km 60% EV Penetration Cable upgrade : Complete replacement of cable with 400mm 2 90% EV Penetration Need complete reconfiguration of the grid Figure 4:Cable details of Rindum Mølleby without cable upgrade 21

33 Figure 5:Cable Details of Rindum Mølleby with cable upgrade for 40% EV penetration Transformer upgrade: The results from the EV integration limits for Bogfinkevej conclude that the cables are strong enough to accommodate 42% of EV penetration but the transformer loading restricts it to 26.2%. Therefore an transformer upgradation by replacement of 250kVA with 400kVA can accommodate 42% EV penetration till the grid needs complete upgrade. Rindum Mølleby transformer of 400kVA can accommodate upto 89% of Ev penetration but the cables restrict it. So, transformer upgrade as a single solution is not feasible. 3.5 Conclusion The EV penetration forecast is taken from the Green emotion project Deliverable D4.3 B2[28] Grid Impact studies of electric vehicles. 22

34 The three different penetration profiles are shown in Figure 6. The high profile corresponds to the high forecast for Denmark. The medium profile corresponds to the medium penetration forecast for Denmark and the high penetration forecast for EU. Both are representative of the expected evolution of EVs, based on an assumption of steady technological progress. The low profile corresponds to the low penetration forecast for the EU. From the EV integration limits of the Bogfinkevej suggests that grid upgradation is not required until EV penetration reaches 26.2%. The EV penetration Figure 6: EV penetration forecast levels in Denmark will reach this limit in 2021 if high forecast is considered and in 2028 if medium forecast is considered. For Rindum the grid upgradation has to take place in It can also be said that network reinforcement cannot be a solution for 100% EV integration in cases like Rindum Mølleby where a complete reconfiguration of the feeders is needed. Based upon the feeder lengths, it can also be generalised that distribution network in urban areas have a larger capacity to accommodate EV/PV installations in comparison to rural networks, where the distance between the houses in the feeder is larger. 23

35 4 4 Reactive power control 4.1 Reactive Power and its effects The voltage drop across two terminals is governed by the equation V = (I active + ji reactive )(R + jx)) It can be seen that both active and reactive current are involved in voltage drop across two terminals. The voltage drop can be mitigated by either delivering the reactive power required at the point of load or by reversing the flow of reactive power against the direction of active power. Voltage control by adjusting reactive power flow is common is high-voltage transmission systems and also in medium voltage distribution systems due to the fact that HV transmission lines have higher reactance than compared to resistance per unit length. Let us consider a circuit as shown in the figure.7, E a be the sending end voltage, V a be the receiving end voltage, R a, X s be the resistance and the reactance of the line respectively. Figure 7:Circuit with a resistance and reactance to compare the sending end voltage and recieving end voltage To study the impact of reactive power on the voltage drop, let s consider three cases where the load consumes reactive power (lagging power factor), neither absorbs nor produces reactive power (unity power factor) and last case in which the load produces reactive power (leading power factor). When the load operates at lagging power factor and consumes reactive power, the additional flow of reactive power causes higher drop in voltage compared to unity power factor where the reactive power doesn t flow. When the load operates in leading power factor, the receiving end voltage can be higher than the sending end voltage due to overcompensation of voltage drop due to the reverse flow of reactive power. This compensation of reactive power at the point of load or reverse flow of reactive power can be used to improve the voltage in a transmission system. 24

36 Figure 8: voltage variation with change in power factor Impact of Cable impedance on reactive power control Low-Voltage distribution systems in the EU have underground cables that are predominantly ohmic when compared to their per unit reactance. Thus the impact reactive power has on the voltage drop across a LV cable is less compared to HV system. The cables in the Bogfinkevej and Rindum Mølleby distribution system have a cross section area of 50mm 2, 95mm 2 and 150mm 2. These have a R/X ratio ranging from 2.5 to 7, so one has to take care that the reactive power is used efficiently without overloading the cables. The R/X ratios of cables used in the distribution system are listed. Table 9: Cable types with their R/X ratios Figure 9: Circuit for checkingvoltage variation of different cable types with p.f. To check the impact of the cable ratios on reactive power control for voltage support, we consider a simple circuit with different cable types of length 500m from the source to the load. A load of sufficient magnitude is selected to get a voltage drop of 5% and the variation in voltage due to change in power factor is plotted for different cable types. Cable type R/X ratio 400 mm mm mm mm mm 2 7 It can be seen form the voltage vs power factor graph that cables with lower R/X ratios have a larger impact with reactive power control. With a high R/X ratio of 4.25 (95mm 2 cable), the voltage improves by 0.01p.u. and even this small improvement could be useful in some cases. Figure 10: Graph showing the variation of Voltage with different cable types 25

37 4.2 PV inverters operating in VAr mode PV inverters have the ability to control reactive power but this is not a standard feature for residential PV inverters, as at present regulations, like German grid code VDE 4105, suggest reactive power control only for PV systems with a peak output greater than 30kWp and systems below 30kWp are operated at unity power factor. On an average PV inverters interact with the grid only for 6-10 hours a day during PV generation, thus limiting its utilization and return on investment of the inverter. The utilization of PV inverters can be increased by operating them in reactive power compensation mode when the PV power is small. During PV generation the inverter consumes some active PV power as losses and for powering the control circuitry. In the absence of PV power, the inverter has to draw some active power from the grid to be able to regulate the DC bus voltage, compensate for switching losses and inject the desired level of reactive power into the grid. This reactive power control capability of the PV inverter during the evening coincides with the tendency of the EV user to charge his vehicle in the evening after returning from work. By using PV inverter for reactive power control, we eliminate the need for oversizing the EV charger to accommodate variable power factor and it can also be considered as a possible solution that can be replicated for large domestic loads such as heat pumps PV Inverter Topology A normal grid connected PV system consists of a PV modules, DC-DC converter, inverter with a DC coupling capacitor and grid connection as represented in the figure. In the absence of PV power, the inverter can be operated in reactive power mode by decoupling the DC-DC converter and the PV module as shown in the figure. Operating the inverter in VAr mode involves the following steps 5 : Pre-charging the DC bus capacitance Regulating the DC bus voltage within limits while regulating the injected reactive power. Figure 11: schematic of inverter operating in VAr mode Pre-charging of the capacitor can be done through either a separate DC connection from an external source or by operating the inverter as a line rectifier. The paper by Maknouninejad [29] studies and explains the use of inverter switches/anti parallel diodes to charge the capacitor and provides design equations for maximum current flowing during capacitor charging. 26

38 4.2.2 Control of Active and reactive power of Inverter The inverter circuit is depicted in the figure12 The inverter measures the DC link Voltage Voltage V dc, the grid voltage v a,v b, v c, the inverter current i a, i b, i c, and the load current i La, i Lb, i Lc. The DC-DC converter is used to boost/buck the operating voltage to the maximum power point tracking for the PV array output. Figure 12: Grid connected inverter circuit The a,b,c voltage and current of the three phase system measured is converted into dq coordinate axis according to the theory of instantaneous reactive power and the theory of the Park transform 6. The relationship of the currents in the two different coordinates is given by the following equation: idq0=ciabc, iabc=c -1 idq0. where idq0=[id, iq, i0] T, iabc=[ia, ib, ic] T, and C is the park transform matrix. C = 2/3 [ cos t Cos( t 2 ) 3 2 Cos( t + 3 ) Sin t Sin( t 2 ) 3 2 Sin( t + 3 ) 1/ 2 1/ 2 1/ 2 ] Where, is the angle frequency of the grid voltage and C -1 =C T because C is an orthogonal matrix. By adjusting the amplitude and the phase of the output voltage of the inverter, the value of the active and reactive power and their direction can be controlled. When the output voltage of the inverter is higher than that of the grid, the inverter outputs inductive reactive power and when the output voltage of the inverter is lower than that of the grid, the inverter outputs the capacitive reactive power 7. 27

39 The equivalent circuit and the and the vector diagram of the single phase inverter is shown in figure 13. In the equivalent circuit, L is the filtering inductor between the grid and the inverter, V 0 is the output voltage, I is the output current of the inverter, V g is the grid voltage and V l is the inductor voltage. In the vector diagram, I d and I q are the active and reactive current output of the inverter, I load is the current of the load, I q = -I load and is the angle between V o and V g. Figure 13: equivalent circuit and phasor diagram of single phase inverter The vectors x represent I, I d, I q, V l, V g and V 0 when active power output increases and reactive power remains constant. It can be seen that the amplitude and the lead forward phase angle of the output voltage increases when active power output is increased and reactive power is kept constant. When the reactive power is increases and the active power output is kept constant, the amplitude increases but the lead forward phase angle decreases. 4.3 Impact of reactive power on EV charging The impact of Reactive power flow in increasing the EV integration limits is studied in the same manner that was used for getting the EV integration limits of the existing grid and reinforced grid. EV s are placed progressively at each house from the transformer end of the feeder to get the best possible case and from the feeder far end to get the worst case scenario. The EV integration limits are calculated by assuming that the EV charger operates at unity power factor and consumes only active power. The reactive power is generated by the PV inverter that should have been idle due to lack of PV generation during the night. The reactive power generated flows opposite to the direction of active power and reduces the voltage drop in comparison to unidirectional flow of active power Bogfinkevej distribution grid The Bogfinkevej distribution system as mentioned earlier was built in 1970 s and supplies to 141 houses. The feeder details are mentioned again for quick reference. 28

40 It can be seen that the distribution system has a majority of cable sized 50mm 2 which provides the least voltage variation with change in power factor. Therefore it is a good example to see if reactive power control can be good solution for integration of electric vehicles. Table 10: Bogfinkevj grid data Bogfinkevej Construction 1970 s Transformer 250kVA Cables (3-phase) 150mm 2 : 0.21km 95mm 2 : 0.51km 50mm 2 : 1.94km Distance between the Feeder 1 : 0.41km substation and last house Feeder 2 : 0.38km connected to the feeder Feeder3 : 0.38km Feeder4 : 0.26km Consumption per year 458MWh Nr. of connected houses 141 The existing grid can accommodate 61 (43% of 141 houses) Electric vehicles during peak winter household demand. This is considering only the feeder capacity to maintain the voltage levels above.95p.u. and when the transformer loading is considered, the EV charging is limited to 37EV s or 26% of 141 houses. Therefore, transformer is the first limiting factor for the bogfinkevej distribution system and the voltage drop limits the EV penetration EV integration limits with Reactive power flow With reactive power control, the EV integration limits are significantly improved for the distribution system with 3 out of the 4 feeders being able to accommodate 100% of EV charging and the EV integration limits increases from 31% to 65%. The PV inverters are limited to operate at 0.8 leading p.f. equivalent of the EV charging active power to avoid increased loading of the transformer/cables and losses. Generating 0.8 p.f. equivalent of reactive power would increase the apparent power flowing in the system by 25% (because P=S*cosfi). Table 11: EV integration limits with reactve power control Bogfinkevej distribution system Feeder1 : 32 houses Unity p.f p.f. leading 0.9 p.f. leading 0.85 p.f. Leading 0.8 p.f. Leading From the transformer Feeder far end Feeder 2 : 51 houses 29

41 From the transformer Feeder far end Feeder3 : 32 houses From the transformer Feeder far end Feeder 4 : 26 houses From the transformer Feeder far end It can be seen that reactive power control for EV charging through PV inverters can be a viable solution for Bogfinkevej in comparison to traditional grid upgrade. In traditional grid upgrade both the transformer and the cables need to be replaced to achieve 90% EV penetration whereas through reactive power control only the transformer needs to be replaced. The comparison between traditional grid upgrade and reactive power control to increase the EV penetration level from the present 26% to 90% presented in the table below: Table 12: Comparison of reactive power control and traditional grid upgrade Reactive Power Control Traditional grid upgrade 40% EV penetration Transformer upgrade: 400kVA Transformer upgrade : 400kVA 60% EV penetration Transformer upgrade: 500kVA Transformer upgrade: 500kVA Cable upgrade: 340m out of 2.72km 90% Ev penetration Transformer upgrade: 750kVA Cable upgrade 54m out of 2.72km Transformer upgrade:600kva Cable upgrade: 950m out of 2.72km Cost implication The use of reactive power for mitigating voltage problems effectively reduces the need for reinforcing the cables in the distribution system. It can be noticed from the above table that in case of 60 % EV penetration, 340m of cable need not be reinforced if reactive power is used to mitigate the voltage problems. The same is true for 90% EV penetration where 900m of cable need not be reinforced. 30

42 Voltage Voltage Bogfinkevej feeder 1 Bogfinkevej Feeeder 1 near the transformer Feeder 1 Bogfinkevej Far end of the transformer Number of Ev's Number of EV's Voltage@ Voltage@.9 The graphs represent the lowest voltage recorded in the feeder 1 when EV are placed progressively from the transformer end of the feeder and from the feeder far end. It can be seen that the voltage improves It can be seen that in both the best case scenario and the worst case scenario that the feeder 1 experiences voltage problems. The profile can be improved by replacing the 150mm 2 cable connecting the transformer and the first house by a cable with a lower R/X ratio and thicker cable as done in the case of Rindum Mølleby which will be discussed later. 31

43 Bogfinkevej feeder 2 It can be observed from the graphs that 100% EV penetration is possible with PV inverters generating 0.9p.f. equivalent reactive power for 3.7kW active power. Feeder2 consists of 150mm 2, 95mm 2 and 50mm 2 cables with a majority of them being 50mm 2 cables. It can also be seen that the lowest voltage with 21 EV s is lower than the lowest voltage with 40 EV s due to over-compensation of reactive power. So, it can be said that in some cases the voltage profile improves with more households charging their EV s and taking part in reactive power control. 32

44 Bogfinkevej feeder 3 Feeder 3 can accommodate 100% EV charging without any voltage violations when PV inverters are producing 0.8.pf. equivalent reactive power for active charging power. Unlike feeder 1 and 2 of Bogfinkevej distribution system, feeder 3 doesn t have any cable with a rating of 150mm 2 and consists of cable with sizes of 95 mm 2 and 50 mm 2. Thus it can be said that for feeders with sufficient capacity, reactive power control can be an effective solution even when the cables have a higher resistance than reactance. 33

45 Bogfinkevej feeder 4 Feeder 4 is a relatively short and is a high EV charging capacity feeder with the last house connected to the transformer being 0.26km away and having an initial EV penetration limit of 76%. All the cables in feeder 4 are of 50mm 2 type and can accommodate 100% Ev charging with just 0.95 leading p.f.. It reiterates the observation from feeder 3 that, for feeders with sufficient capacity, reactive power control can be an effective solution even when the cables have a higher resistance than reactance 34

46 4.3.2 Rindum Mølleby Distribution grid The relatively new distribution grid of Rindum Mølleby was constructed around the year 2000 and its details are mentioned in the table. Table 13: rindum grid summary The Rindum Mølleby distribution grid is mostly made up of cable of 150mm 2 size and should be better suitable for reactive power control compared to Bogfinkevej, but the length of the feeders in a major hinderance. Rindum Mølleby Construction 2000 s Transformer 400kVA Cables (3-phase) 150mm 2 : 2.55km 95mm 2 : 0.55km Distance between the Feeder1 : 1.04km substation and last house Feeder2 : 0.546km connected to the feeder Feeder3 : 0.687km The existing grid can accommodate 22 Consumption per year 399 MWh EV s before experiencing voltage Nr. of connected houses 93 problems. When only the transformer capacity is considered, the distribution system can accommodate an EV penetration limit of 89%. Thus it can be said that voltage drop due to the long distance of the feeders is the first limitation for the feeder and then the transformer loading acts as a limitation EV integration limits with reactive power flow The EV integration limits with reactive power flow is not significantly different when compared to the EV integration limits without reactive power. Feeder 1 and feeder3 show no improvement or ne gligible improvement due to their long distance between the transformer and the first house. Feeder 2 which is shortest and has the highest capacity with an existing EV integration limit of 19 EV s (73%) can attain 100% EV integration limit with EV s being charged at an equivalent leading power factor of 0.9. Table 14: EV integration limits with reactive power flow Rindum Mølleby distribution system Feeder1 : 34 houses Unity p.f p.f. leading 0.9 p.f. leading 0.85 p.f. Leading 0.8 p.f. Leading From the transformer Feeder far end Feeder 2 :19 houses From the transformer Feeder far end Feeder3 : 40 houses From the transformer

47 Feeder far end Rindum Mølleby distribution grid with upgraded Cables The existing Rindum Mølleby Distribution grid cannot integrate EV s in their feeders1 and 3 with reactive power flow. To improve its response to reactive power, we replace the existing cables between the transformer and the first house with 400mm 2 cable which not only decrease the voltage drop due to active power but also has a lower R/X ratio of 0.68 compared to R/X ratio of the existing 150mm 2 cable which is 2.5. Table 15: EV integration limits with cable upgraded Rindum Mølleby modified distribution system Feeder1 : 34 houses Unity p.f p.f. leading 0.9 p.f. leading 0.85 p.f. Leading 0.8 p.f. Leading From the transformer Feeder far end Feeder 2 :19 houses From the transformer Feeder far end Feeder3 : 40 houses From the transformer Feeder far end It can be seen that by replacing the existing cables of length 0.64km in feeder 1 and 0.27km in feeder 3, we can achieve 100 % EV in both the feeders. The comparison between traditional grid upgrade and reactive power control to increase the EV penetration level from the present 23% to 90% presented in the table below: Table 16: Comparison of reactive power control with traditional grid upgrade Reactive power Control Traditional grid upgrade 40% EV Penetration Cable upgrade: 0.91 km from Cable upgrade : 1.7km from 2.9km 2.9km 60% EV Penetration No upgrade needed Cable upgrade : Complete replacement of cable with 400mm 2 90% EV Penetration Transformer upgrade to 600KVA Need complete reconfiguration of the grid 36

48 For distribution systems like Rindum Mølleby, where either reactive power or traditional grid upgrade cannot be a complete solution, a combination of both is a viable solution Rindum Mølleby distribution grid with reconfigured transformer position The Rindum Mølleby distribution grid like mentioned before has long cable from the transformer to the first house. The distance between the transformer and the house is reduced to 1m and simulated using reactive power flow like it was done in the previous chapter. Table 17: Rindum EV integration limits with reconfigured transformer position Rindum Mølleby modified distribution system Feeder1 : 34 houses Unity p.f p.f. leading 0.9 p.f. leading 0.85 p.f. Leading 0.8 p.f. Leading From the transformer Feeder far end Feeder 2 :19 houses From the transformer Feeder far end Feeder3 : 40 houses From the transformer Feeder far end Like observed from the results in the previous chapter, at unity power factor feeder2 & 3 can accommodate 100% EV penetration but feeder1 can charge 10 EV s in the worst case. This performance of the feeder can be improved by using reactive power. When the EV is being charged at an effective power factor of 0.8 leading, the feeder can accommodate 17 EV s ie.50% EV penetration. 37

49 Rindum Mølleby feeder 1 This graph compares the combination of solution used for improving the EV penetration limits of the Rindum Mølleby feeder 1. It can be observed that the feeder lowest voltage when 1 EV is placed is around due to the long cable between the transformer and the first house. This voltage drop is reduced when the cable is either replaced or the transformer is placed 1m away for the first house. It can also be observed that the voltage profile in case of the cable being upgraded or the transformer being shifted is almost the same. But when reactive power flows, the voltage improvement in case of cable upgrade is better due to lower R/X ratio of 400mm 2 cable. 38

50 Rindum Mølleby feeder 2 It can be seen that feeder 2 of Rindum Mølleby can accommodate 100% EV integration when the EV charger is operating at an equivalent leading power factor of 0.9. Unlike feeder 1 and feeder 3, combination of solutions are not required to get 100% EV integration and it again reiterates the observations made in Bogfinkevej feeders that reactive power control can be viable option on its own when the feeders are short and have sufficient initial capacity to accommodate EV charging. 39

51 Rindum Mølleby feeder 3 This graph like the Rindum Mølleby feeder 1 graph compares the different combination of solutions for integration of EV s. Reactive power control alone cannot support voltage, So additional support in the form of cable upgrade and transformer relocation is considered. It is observed that transformer shifting alone can accommodate 100% EV charging unlike in the case of feeder1. With the upgraded cable the voltage profile improves but needs the help of reactive power to get be able to charge 40 EV simultaneously. 40

52 4.4 Yearly simulation of Distribution system with high PV and high EV penetration Bogfinkevej Distribution grid The Bogfinkevej distribution system is simulated for a complete year with a 60% penetration of EV s and photovoltaic systems. The EV s charging from the residential connection consumes 3.7kW and the peak output from the PV system is considered to be 5kW. The distribution system is first simulated with the existing grid and without reactive power, and the results are then compared with the grid with an upgraded transformer and with reactive power control. Voltage support through reactive power control is done by operating the EV charger at an equivalent power factor of 0.8 leading for feeders 1,2,3 and at 0.95leading for feeder 4 based on their existing capacity to integrate EV s. The distribution system has a cumulative peak PV output of 420kW and it experiences over-voltage s when the cumulative PV production from the system exceeds 250kW. In the complete year the net PV production exceeds 250kW in only 665 hours in a year. This is due to self-consumption and the simultaneity factor of the systems generating power. During instances of PV production exceeding 250kW, the PV systems are made to operate at 0.8p.f. lagging. The results show that EV s and PV systems can be integrated into the system effectively by using reactive power but, at the cost of higher losses due to the increased flow of reactive power as seen in table 19. If the system operates at a power factor of 0.8, then the apparent power increases by 25%. Due to the increase in apparent power, the I 2 R losses increase by 56%. Apparent power(s) = Active power (P) 2 + Reactive power(q) 2 Activepower (P) Apparent power (S) = Cos(powerfactor(φ)) The table below shows the energy consumed and exchanged through the distribution system, for calculating the change in losses and efficiency due to reactive power. Table 18: Energy Exchange between the distribution grid and the consumer MWhr for year 2012 Household demand (a) 444 Electric Vehicle charging demand (b) 473 Total PV generation (c) 486 Self-consumption by houses having PV (d) 129 Energy taken from the grid (e)=(a + b - d) 788 Energy fed into the grid (f) =(c - d) 357 Energy exchange with the grid (e + f)

53 Table 19: Losses comparision with different Solutions Original grid Grid with reactive power flow Grid with cable upgrade Losses 23.41MWhr 35.27MWhr 19.60MWhr Diff. in losses +50.6% -16.2% Efficiency 97.9% 96.91% 98.2% 42

54 43

55 44

56 45

57 46

58 4.4.2 Rindum Mølleby Distribution grid The Rindum Mølleby distribution grid is simulated for an entire year like the Bogfinkevej distribution system to calculate the losses and compare the solutions. One of the solution is by reconfiguring the transformer near the feeder and the other solution is by using reactive power control in combination with upgrading the cable with 400mm 2 cable. By replacing the cable with a 400mm 2 cable, the voltage profile along the long cable improves and its response to reactive power increases when compared to 150mm 2 cable, as seen in the section From the graphs below it can be seen that with the existing cables and transformer position, the lowest voltage during EV charging is p.u. and is improved by using reactive power and transformer repositioning. The losses in case of both the solutions are compared in the table below. It can be noticed that the increase in losses due to reactive power flow is not as significant as in the case of Bogfinkevej distribution system due to the decreased losses in the upgraded cable. When we compare the losses with and without reactive power in the grid with upgraded cables, we get 35% increase in losses due to reactive power, which is comparable to Bogfinkevej distribution system case. Table 20:Energy Exchange between the distribution grid and the consumer MWh for year 2012 House hold demand (a) 399 Electric Vehicle charging demand (b) 312 Total PV generation (c) 316 Self-consumption by houses having PV (d) 110 Energy taken from the grid (e)=(a + b - d) 601 Energy fed from the grid (f) =(c - d) 206 Energy exchanged with the grid (e + f) 807 Table 21: Losses comparision with different Solutions Original grid Grid with reactive power flow and upgraded cables Grid with upgraded cables Losses MWhr MWhr 9.78 MWhr Diff. in losses % -34.8% Efficiency 98.14% 98.08% 98.7% 47

59 Original grid Grid with upgraded cables and reactive power 48

60 49