Distributed Power Conversion Architecture for Microgrids and Integration of Renewable Energy Sources

Transcription

1 Distributed Power Conversion Architecture for Microgrids and Integration of Renewable Energy Sources by Junjian Zhao A thesis submitted in partial fulfillment of the requirement for the degree of Master of Science (Electrical and Computer Engineering) at the UNIVERSITY OF WISCONSIN-MADISON 2013

2 Copyright by Junjian Zhao 2013 All Rights Reserved

3 Distributed Power Conversion Architecture for Microgrids and Integration of Renewable Energy Sources by Junjian Zhao Under the supervision of Professor Yehui Han at the University of Wisconsin Madison Approved by: Yehui Han Date:

4

5 ABSTRACT This article focuses on distributed power conversion architecture for microgrid and integration of renewable energy sources from the aspect of power electronics converter. It covers the following areas: Power Distribution and Control, Alternative Power and Energy Technology, Energy Storage Technology, Energy Conversion and Efficiencies. The state-of-the-art research of microgrid, control algorithms and power electronics building blocks are reviewed in this article. We propose a distributed power conversion architecture for renewable energy sources in microgrids, which has advantages including functionality, efficiency, reliability, and low manufacturing cost. Its impact on future grids is investigated. The physical system in University of Wisconsin-Milwaukee is studied. A family of DAB converter based power electronics building block is proposed to realize the modular design consideration. The modeling and control algorithms of the DC and AC power electronics building blocks are developed and simulation results have verified the theoretical model. I

6 This research also opens up new possibilities for how renewable energy sources can be integrated and applied. Keywords: Renewable Energy Power Conversion Microgrid Power Electronics Building Blocks II

7 ACKNOWLEDGMENTS First and foremost, I would like to extend my deep appreciation towards my advisor, Professor Yehui Han, for his generous support and guidance during the project. His experience in the field of power electronics, and his willingness to share it with his colleagues and students, remain as a great source of inspiration to me. Secondly, I feel indebted to all other faculty members of Wisconsin Electric Machinery and Power Electronics Consortium (WEMPEC). Their lectures furthered my understanding of machines and power electronics and built the foundation of my research. I gratefully acknowledge Wisconsin Energy Research Consortium (WERC) for the financial support of this project. I would like to thank Prof. Yu s group, especially Carl and Qiang from University of Wisconsin-Milwaukee for their work from power system side and preparing the final report of this project. I appreciate the help of Liang Wang from Tsinghua University, he helped to improve converter control algorithms. III

8 I would like to thank all my WEMPEC colleagues, especially Jiyao, Ye, Bo, Tim, Kenton, Silong, Joyce, Di, Yingjie, Minjie, and Yukai for making my life in Madison enjoyable and meaningful. Finally, I would like to thank my parents and my girlfriend Emily X. Zhang for their love and support all these years. IV

9 TABLE OF CONTENTS ABSTRACT...I ACKNOWLEDGMENTS...III TABLE OF CONTENTS... V LIST OF FIGURES... IX LIST OF TABLES... XIII Chapter 1 Background The picture of a Microgrid Distributed Generators Energy Storage Devices Distributed Power Conversion Architecture Microgrid Loads Technical Challenges for Power Conversion in Microgrids Outline of the Thesis Chapter 2 State of the Art Review System Level - Microgrid State of the Art AC Microgrid and Applications DC Microgrid and Applications Hybrid AC/DC Microgrid and Applications Control Level - Droop Control in Microgrids State of the Art Droop Control in AC Microgrids Droop Control in DC Microgrids Droop Control in Hybrid AC/DC Microgrids Converter Level - Power Electronics Building Block (PEBB) State of the Art The Concept of PEBB Plug and Play V

10 2.3.3 Modeling and Hierarchical Control of PEBB Chapter 3 PEBB Based Power Conversion Architecture for DC Microgrids Proposed Architecture Realization in UW-Milwaukee Microgrid DC Bus Voltage Level DC Bus Voltage Identification The Advantages of 380V DC Bus Voltage PEBB Topology Selection Bidirectional Dual Active Bridge The overview of PEBB Topology Candidates Single-Stage Topologies Two-Stage Topologies PEBB Topology Identification Modular Design Considerations Chapter 4 Modeling of DAB Based PEBBs Modeling of DAB based DC/DC PEBBs Lossless Model of Single Phase DC/DC DAB Average Model of Single Phase DC/DC DAB Small Signal Model of Single Phase DC/DC DAB Three Phase DAB based DC/DC PEBBs Modeling of DAB based AC/DC PEBBs Solution 1: Combination of Traditional AC/DC topology and DC/DC DAB Solution 2: Pure Single Phase AC/DC DAB Chapter 5 Control Algorithms and Simulation of PEBBs Control Method for Single Phase DC/DC DAB Phase Shift Modulation Closed-loop Control Control Method for Traditional Three Phase AC/DC Converter Control Method for Single Phase AC/DC DAB Control Method for Three Phase AC/DC DAB VI

11 5.5 Simulation Results Simulation of DC/DC Single Phase DAB Simulation of AC/DC Single Phase DAB Simulation of Traditional Three Phase AC/DC Converter Simulation of Three Phase AC/DC DAB Chapter 6 Conclusion and Future Work References VII

12 VIII

13 LIST OF FIGURES Figure 1.1 Centralized utility vs. distributed generator-based utility...3 Figure 1.2 Global greenhouse gas (GHG) abatement cost curve 2030 [3]4 Figure 1.3 Centralized power conversion architecture Figure 2.1 CERTS Microgrid Architecture Figure 2.2 Real power droop curve for power sharing Figure 2.3 University of Wisconsin-Madison Microgrid Architecture Figure 2.4 EPRI 380V-DC data center distribution [14] Figure 2.5 (a) Typical AC distribution vs. (b) Standard 380V DC distribution [14] Figure 2.6 An example of hybrid AC/DC microgrid Figure 2.7 Typical hybrid AC/DC microgrid structure Figure 2.8 Active and reactive power droop characteristic for AC microgrid Figure 2.9 Active power droop characteristic for DC microgrid Figure 2.10 Power Electronics Building Block Figure 2.11 PEBB concept for power electronics Figure 3.1 Proposed distributed power conversion architecture Figure 3.2 University of Wisconsin-Milwaukee microgrid map Figure 3.3 DC bus voltage selection overview Figure 3.4 Comparison of energy usage for different system voltages.. 58 Figure 3.5 Comparison of conductor diameters (courtesy of UECorp) [14] Figure 3.6 General structure of an isolated, bidirectional DC/DC converter Figure 3.7 Family tree of single-stage and two-stage bidirectional isolated DC/DC topologies Figure 3.8 Single phase DAB converter Figure 3.9 Three phase DAB converter Figure 3.10 Voltage sourced full bridge with a current sourced full bridge IX

14 converter Figure 3.11 Voltage sourced full bridge with a current doubler Figure 3.12 Voltage sourced full bridge with a push-pull structure on low voltage side Figure 3.13 Bidirectional series-parallel resonant LLC converter Figure 3.14 Bidirectional series-parallel resonant LCC converter Figure 3.15 Two stage converter with buck converter on high voltage side Figure 3.16 DC/DC PEBB modular Figure 3.17 AC/DC PEBB modular Figure 3.18 AC/AC PEBB modular Figure 4.1 Single phase DC/DC DAB Figure 4.2 Single phase DC/DC DAB lossless model Figure 4.3 Traditional three phase AC/DC converter Figure 4.4 Single phase AC/DC DAB Figure 4.5 Single phase AC/DC DAB based three phase AC/DC PEBB 90 Figure 5.1 Control schematic of single phase DC/DC DAB Figure 5.2 Control block diagram of single phase DC/DC DAB Figure 5.3 Block diagram of dq frame decouple control for traditional three phase AC/DC converter Figure 5.4 Control block diagram of traditional AC/DC converter Figure 5.5 Control algorithm of single phase AC/DC DAB Figure 5.6 Block diagram of single phase AC/DC DAB with controller100 Figure 5.7 Control method for three phase AC/DC DAB Figure 5.8 Block diagram of three phase AC/DC DAB with controller102 Figure 5.9 Simulink model of single phase DC/DC DAB based PEBB 103 Figure 5.10 Simulink model of phase shift modulation signal generator for DC/DC DAB Figure 5.11 Simulink model of single phase DC/DC DAB with controller and mathematical counterpart Figure 5.12 Simulation result of output voltages of detailed DAB model and X

15 linearized model Figure 5.13 Detailed view of output voltages of DC/DC DAB during source and load disturbance Figure 5.14 Detailed view of output voltages of DC/DC DAB during start period Figure 5.15 Algorithm for obtaining d for AC/DC DAB Figure 5.16 Simulink model of single phase AC/DC DAB based PEBB and phase shift modulation signal generator Figure 5.17 Simulink model of phase shift modulation signal generator for DC/DC DAB Figure 5.18 Waveform of LC filter Figure 5.19 Simulink model of traditional three phase AC/DC converter113 Figure 5.20 Mathematical model of traditional three phase AC/DC converter Figure 5.21 Mathematical model of the controller traditional three phase AC/DC converter Figure 5.22 Traditional three phase AC/DC converter with controller and mathematical counterpart (part 1) Figure 5.23 Traditional three phase AC/DC converter with controller and mathematical counterpart (part 2) Figure 5.24 Simulation result of output voltages of detailed traditional three phase AC/DC converter model and linearized model Figure 5.25 Detailed view of output voltages of detailed traditional three phase AC/DC converter model and linearized model during disturbance Figure 5.26 Detailed view of output voltages of detailed traditional three phase AC/DC converter model and linearized model during start up Figure 5.27 Simulation result of dq current of detailed traditional three phase AC/DC converter model and linearized model Figure 5.28 Detailed view of dq current of detailed traditional three phase AC/DC converter model and linearized model during disturbance Figure 5.29 Detailed view of dq current of detailed traditional three phase AC/DC converter model and linearized model during start up Figure 5.30 Simulink model of three phase AC/DC DAB with controller125 XI

16 Figure 5.31 Simulation result of output voltages of detailed three phase AC/DC DAB model and linearized model Figure 5.32 Detailed view of output voltages of detailed three phase AC/DC DAB model and linearized model during disturbance Figure 5.33 Detailed view of output voltages of detailed three phase AC/DC DAB model and linearized model during start up Figure 5.34 Voltage and current waveforms after low-pass filter XII

17 LIST OF TABLES Table 1.1 Emission and Efficiency Comparison w/ and w/o CHP...5 Table 1.2 Typical interfaces used with DER...8 Table 3.1 Summary of loading and solar capacity per building Table 3.2 Bus summary of loading and solar capacity Table 3.3 Detailed specifications of power converters XIII

18 XIV

19 Chapter 1 Background Conservation of energy resources, environmental protection and sustainable development have been viewed as three most important challenges for decades [1]. The growing demand of energy needs urges collective and long-term efforts on energy development with an emphasis on protecting natural resources and the environment. Building upon the existing electrical infrastructure, the use of low cost renewable energy sources for electric power generation seems to be an effective and reasonable approach to address the challenges we face. Distributed power conversion architecture and the use of microgrids represent better alternatives compared to centralized generation especially when renewable energy sources are located close to end users. 1.1 The picture of a Microgrid Microgrids are emerging as a critical feature of the future smart-grid based power systems. A microgrid is constituted of three major components: integrating loads, distributed generators (DG) and energy storage devices [2]. Renewable energy sources can also be integrated into various types of microgrid. As an attractive approach for addressing the growing demand of energy, the role of 1

20 renewable energy sources is unique and important. Microgrids can operate either in parallel with the grid, or as an autonomous and independent power island; it could also be in a transition between grid-connected mode and islanded mode of operation [2] Distributed Generators Centralized electricity generation, transmission and distribution in Figure 1.1(a) has dominated almost all large integrated power systems in the world such as large-scale hydro, coal, natural gas and nuclear power plants. They have been used for decades and seem to be the only resources which power generation could rely on. However, the centralized solution has numerous drawbacks comparing to the merging distributed DG-based power system which is shown in Figure 1.1(b), including long-distance between supply (natural resources) and demand (where people live and factories locate), lossy high-voltage transmission carried the power to the customers from centralized sources. Moreover, the distributed power systems which are enabled by DGs, especially the DG-based microgrids have met a growing demand for clean, reliable and affordable electricity generation, which have constituted a huge driving force on changing the current power generation situation which heavily relies on centralized power generation, transmission and distribution architecture in to a DG-based new generation. 2

21 CONVENTIONAL UTILITY Power Plant (~ 1000MW) Transmission Line (~ 100 miles) Distribution Line (~ 10 miles) Substation Distribution Transformer (a) Conventional centralized utility DISTRIBUTED GENENRATOR BASED DISRIBUTION Power Plant (~ 1000MW) Robust Generation and Distribution Transmission Line (~ 100 miles) Distribution Line (~ 10 miles) M Substation Distribution Transformer Distributed Utility Provides Local Reliability, Intentional Islanding and CHP (b) DG-based utility Figure 1.1 Centralized utility vs. distributed generator-based utility 3

22 There are some well-established generation technologies, for instance, induction generators and synchronous generators; as well as emerging technologies, such as combined heat and power (CHP, also known as cogeneration), fuel cells, wind turbines, photovoltaic (PV), micro-turbines can be used as the generation technologies applicable for a microgrid. Among the distributed generators, the fuel cell, micro-turbines and internal combustion-engine generator belong to dispachable sources due to the mature control techniques for these kinds of DGs; whereas wind turbines and PV belong to intermittent sources, since the wind speed and solar irradiance are hard to predict. Nevertheless, it is important to note that CHP, wind power generation and photovoltaic have shown a rapid growth in technological research and applied industry products in past two decades. Figure 1.2 Global greenhouse gas (GHG) abatement cost curve 2030 [3] 4

23 The economic costs of reducing carbon emissions are illustrated in Figure 1.2. It indicated that industrial efficiency improvements as CHP had been found to make noticeable improvements in carbon emissions while at the same time making profits for users because they were negative costs in the chart. Table 1.1 showed the emission and efficiency comparison with and without CHP. Table 1.1 Emission and Efficiency Comparison w/ and w/o CHP Energy source type NO x CO 2 Efficiency (lb./kwh) Micro-turbine (0.0004) 1.19 (0.45) 30% (~80%) Gas Turbine (0.001) 1.15 (0.50) 35% (~80%) Fuel cell (0.45) 38% (~80%) Gas Adv. Engine (0.0024) 0.94 (0.45) 38% (~80%) ( ) indicates use of waste heat: combined heat and power (CHP) Air pollution emission impacts associated with economic market potential of DG in California, June 2000 Renewable energy sources like wind and PV are also able to reduce a large amount of carbon emissions; however they require huge costs to implement. Different microgrid technologies serve as a platform upon which CHP systems, PV and 5

24 wind power systems can be integrated, hence, microgrid technologies are particularly attractive in the near future. Given the advantages of microgrid technologies certain renewable energy sources and CHP systems can operate in either a grid-tied or a stand-alone mode, or it could function as part of a microgrid Energy Storage Devices Energy storage devices, a major component of microgrid, are very critical for successful operation of microgrids. Taking care of balancing the power and energy demand with generation is the main function of the energy storage devices in a typical microgrid application [4]. Energy storage devices take this responsibility in three different approaches: 1. Insure the power balance in a microgrid for the purpose of avoiding load fluctuations as well as DGs improper responses to these disturbances because of their lower inertia. 2. When there are dynamic variations in intermittent energy sources, energy story devices provide ride-through capability and allows the DGs to operate as dispatchable units. 3. Provides the initial energy requirement for a seamless transition between grid-connected to/from islanded operation of microgrids. Among the available energy storage technologies [5] [6], batteries, fly-wheels and super-capacitors seem to be more applicable for microgrid type of setup [7]. In the use 6

25 of a flywheel, it can be used as a central storage system for the whole microgrid. In the use of batteries, either storage can be mounted on the DC bus of each micro-source or can be used as a central storage system. Batteries provide extra function being able to reserve energy for future demand. Super capacitors would be an expensive choice compared to both batteries and flywheels [5]. Another option is to have a large traditional generation having considerable inertia along with the micro-sources in the system [4] Distributed Power Conversion Architecture There are a variety of distributed power conversion architectures that could be used for microgrids and renewable energy systems. They can be classified as direct current (DC), alternating current (AC), and hybrid of DC and AC. The terminology of distributed power conversion architectures refers to the hierarchy and interface of distributed energy resources (DER), representing both DG and energy storage technologies. Most of the emerging DER technologies require an inverter interface for the purpose of converting the energy into grid-compatible AC power. This interfacing consists of both DC/DC converter and DC/AC inverter. It could also be made up of only an inverter [4]. The power electronics interface will be accommodated with filters, control blocks and necessary protection systems. The converter s capability of voltage and frequency 7

26 control allow these DER units to support the microgrid operation. Table 1.2 summarizes the interfacing and power flow control options of common DER. Table 1.2 Typical interfaces used with DER Primary energy source type Typical interface DG CHP Synchronous generator Internal combustion engine Synchronous or induction generator Small hydro Synchronous or induction generator Fixed speed wind turbine Induction generator Variable speed wind turbine Power electronics converter (AC DC AC) Micro-turbine Power electronics converter (AC DC AC) Photovoltaic (PV) Power electronics converter (DC DC AC) Fuel cell Power electronics converter (DC DC AC) Energy storage Battery Power electronics converter (DC DC AC) Fly-wheel Power electronics converter (AC DC AC) Super capacitor Power electronics converter (DC DC AC) Hence, distributed power conversion architectures based on power electronics are essential for future grids. 8

27 1.1.4 Microgrid Loads A microgrid could serve various customers, such as residential, commercial and industrial. Defined as critical/sensitive loads, commercial and industrial users demand high degree of power quality and reliability, which puts special requests on the microgrid setup in order to achieve the expected operating strategy: 1. Facilitate load/generation shedding within the microgrid in order to meet the net import/export power in grid connected mode. 2. Facilitate load/generation shedding in order to stabilize the voltage and frequency in the autonomous operation. 3. Improve both the power quality and reliability of specified critical and sensitive loads. 4. Balance or reduces the peak load to optimize the ratings of DER. Part of the non-sensitive loads can be used as controllable loads to achieve the above operating strategies in a microgrid [8-10]. 1.2 Technical Challenges for Power Conversion in Microgrids Just as mentioned in section 1.1.3, most of DER requires a power electronics interface in order to be compatible with traditional AC power grid. Nowadays, like the centralized power utility systems which is shown in Fig. 1.1(a), huge numbers of different type of renewable energy sources and energy storage 9

28 devices have been largely integrated into traditional AC power grids by centralized power electronics such as DC/AC converters or/and AC/AC converters, as shown in Fig This kind of structures which are also known as centralized power conversion architectures. Battery Banks PV Wind Power Loads or Diesel 3 phase AC System Centralized DC/AC Converters Figure 1.3 Centralized power conversion architecture However, the centralized power conversion architecture has numerous severe limitations including low efficiency, low reliability, low power density and high cost, which can be elaborated as below: 10

29 1. Centralize converters perform all the functions of power conversion, management and integration. As the number of functions increase, these converters are overworked; as a result, their performance and reliability degrade. 2. Centralized converters are unable to reduce mismatches within solar modules or battery cells. The output capacity of renewable sources cannot be fully utilized. 3. Converters cannot be modularized and mass-produced, accordingly, their manufacturing cost is high. 4. The whole system reliability relies heavily on centralized converters. Any malfunction of a centralized converter could lead to the shutdown of the entire power generation from all the renewable sources that are connected to it. 5. Without a common DC buffer for all the renewable sources, power flow in ac buses (output) would experience a large fluctuation due to the uncertainty power output from renewable energies. 6. Centralized converters often need a line frequency transformer which is heavy and expensive: with an extra cost associated with site preparation and installations of 10, 0000 lbs. of equipment. This architecture seems to be inadequate to satisfy the needs of future grids of being smarter and affordable, as well as of being able to integrate renewable energy sources. 11

30 1.3 Outline of the Thesis In order to better solve this problem, distributed power conversion architecture is proposed. Chapter 2 provides a detailed review of previous research studies relevant to the current study. Their strengths and limitations will be examined, which serves as the basis for the current study. The state-of-the-art of system level research, control level research and converter level research, i.e. microgrid, droop control and power electronics building blocks will be examined. The motivation for using the proposed distributed power conversion architecture for microgrid and integration of renewable energy sources will also be presented. Chapter 3 provides a research approach and the detailed structure of proposed distributed power conversion architecture. The realization of the proposed power conversion architecture in University of Wisconsin-Milwaukee campus microgrid is introduced and the system parameters are also given in this chapter. Proper DC bus voltage which offers a better efficiency and reliability will be discussed also. Moreover, the topology selection of power electronics building blocks for proposed architecture is elaborated also. Possible candidates are listed and examined one by one. Some modular design considerations inspired by power electronics building blocks are given at the end of this chapter. 12

31 Chapter 4 documents the modeling of dual active bridge based power electronics building blocks. Chapter 5 provides the control algorithms and simulation results for the operation of dual active bridge based power electronics building blocks in the proposed distributed power conversion architecture. Chapter 6 summarizes the key conclusions and thesis contributions as well as proposes follow-up research. 13

32 14

33 Chapter 2 State of the Art Review 2.1 System Level - Microgrid State of the Art Microgrids have received a considerable amount of attention in recent years, including several applications installed around the world. Most of the work has been focusing on safe, stable and reliable operation during both grid-tied and stand-alone operation. As identified by existing literature, there are mainly three different categories of architectures for microgrids: AC microgrid, DC microgrid and hybrid AC/DC microgrid. The droop control is the most widely used control method in all types of microgrids AC Microgrid and Applications Thematically speaking, the AC microgrid concept presented by the Consortium for Electric Reliability Technology Solutions (CERTS) is being recognized as an advanced approach for enabling integration of an unlimited quantity of distributed resources into the electricity power grid in a cost-effective manner. The terminology CERTS microgrid [11] indicates a collection of sensitive loads and micro-sources connected to power utility by a static switch which is shown in Figure By using 15

34 micro grids, the quality of power seen from the demand side and the reliability of the supply could be significantly improved. Load Bank 3 And Fault Load Load Bank 4 And Fault Load Converter Based Source Drive by Natural gas 60kW Converter Based Source Drive by Natural gas 60kW Fault Load 2 Static Switch 112kVA 480V/480V Load Bank 5 And Fault Load 500kVA 480V/480V 15MVA 13.2kV/480V Load Bank6 And Fault Load Converter Based Source Drive by Natural gas 60kW Figure 2.1 CERTS Microgrid Architecture 16

35 The major features of the CERTS AC microgrid could be summarized as [11]: 1. Peer-to-peer environment In the CERTS microgrid, there was no central controller, this special characteristic allowed the power sources to be operated in a peer-to-peer way. The peer-to-peer function did not allow any components, like a master controller or central storage unit that was critical or necessary for operation of the microgrid, suggesting that the microgrid was able to continue operating even with loss of any component or generator. Under this circumstance, with one additional source (N+1), the complete functionality of the CERTS AC microgrid could be guaranteed even with the loss of any source. 2. No explicit communication system Combine loads with sources, most current microgrid implementations permitted intentional islanding and attempt to use the available waste heat. These solutions were contingent on complex communication and control. However, CERTS microgrid needed to provides these features without a complex control. Via providing generator-based controls that enabled a plug-and-play model without communication or custom engineering, this approach outweighed CERTS microgrid in terms of simplicity and efficiency. 3. Plug-and-play Plug-and-play was referred to as a unit which could be placed at any point on the electrical system without reengineering the controls. One of the most important 17

36 features of the plug-and-play model was that it facilitated placing generators near the heat loads, therefore allowing more effective use of waste heat without complex heat distribution systems such as steam and chilled water pipes. 4. Integration with CHP to improve efficiency By adopting the unit power control configuration, each DG regulates the voltage would magnitude at the connection point and the power that the source is injecting. This was the power that flowed from the micro source. Employing this configuration means that extra powers from the grid would follow if a load increase anywhere in the microgrid because every unit regulated to constant output power. This configuration fitted CHP applications because production of power relying on the heat demand. It was not until the waste heat had been utilized, that electricity production could be referred as making high efficiencies. The power was balanced within the island when the system islands the local power vs. frequency droop function was ensured. 5. Smooth transfer between island and grid-tied operation The thyristor based static switch was able to autonomously island the MG because of grid congestion or disturbances such as faults, IEEE 1547 events or power quality events. When the established re-connect criteria being satisfied, the reconnection of the MG would be autonomously achieved. Resynchronization was achieved by using the frequency difference between the islanded microgrid and the utility grid. As shown in 18

37 Figure 2.2, each micro source could seamlessly balance using the reactive power versus voltage droop, therefore providing local stability, and at the meantime guaranteed that there were no circulating reactive currents between sources. Frequency (p.u.) Real Power (p.u.) Figure 2.2 Real power droop curve for power sharing [12] demonstrated the control of inverter-based micro sources for operation in a micro grid environment using both real and reactive power droop controllers. It was necessary for static micro sources such as fuel cells and PV to use the inverter-based approach because it would convert the DC voltage to AC voltage acceptable by the electric power grid or local load. For dynamic micro sources such as high-speed permanent magnet synchronous generators, [12] reported the cascaded inverters with 19

38 droop controls. The following section of 2.2 contained more details about the droop control. The CERTS microgrid concept was implemented in a proof-of-concept hardware setup in Wisconsin Electric Machine and Power Electronic Consortium (WEMPEC) lab at the University of Wisconsin-Madison. This Microgrid included two sources, five sets of three phase loads and a static switch to allow connection to the grid. Figure 2.3 showed the component layout. Notice the overall 100 yd. cable between the sources in order to better capture the voltage drops that normally exist on feeders [13]. A custom made board allowing interface with a Digital Signal Processor (DSP) replaced the off-the-shelf inverter s control card. The control was digitally implemented in the DSP to drive the behavior of the power inverter. The measurement of the currents flowing on the local feeder was used in the feeder flow control option. Notice that during feeder flow control, the measure of the currents injected by the unit was also needed to calculate the reactive power and the active power to enforce its limits. 20

39 Utility System 480V 208V Static Switch 4 Wire Cable 4 Wire Cable, 75yd 4 Wire Cable, 25yd 480V Load Load Load 480V MS1 MS2 Figure 2.3 University of Wisconsin-Madison Microgrid Architecture DC Microgrid and Applications Compared to AC microgrids, DC microgrids have not been developing for many decades. DC technology has changed significantly over the last more than one hundred and twenty years since T. E. Edison and Westinghouse publicly battled over DC power grid versus AC power grid in the War of Currents. 21

40 Since transformers could solve the problem of obtaining power more than a kilometer from a centralized power station in a cheap and effective way, AC managed to be the dominant force in power distribution design in the 19th century even with ongoing rapid technology development. However, today ac seems not being able to satisfy the need of market in power distribution of higher efficiency, better reliability, and more intelligence. It has long been recognized that at the end of the power chain is marked by DC loads. Since the 1960s, semiconductors (natively DC) have come to dominate our electrical devices such that the majority of the load base will be natively DC soon. Since most carbon-free energy sources and energy storage systems are natively dc. They are being deployed in a distributed manner. Therefore, assumptions underlying AC power distribution need to be reexamined. In order to achieve DC voltage requirements, the current widely accepted practice of AC power distribution was to add layers of power conversion However, these conversions brought about inefficiencies, possible series fault points, and maintenance complexity in getting from the AC drop to the critical load. Report of Edison Redux [14] looked at low-voltage DC (LVDC) at 380 V, the new industry specification and the single worldwide standard in data centers (and one envisioned for all future commercial buildings as well). A 380Vdc building level 22

41 power grid enabled high efficiency appliances like lighting and supporting the current industry interest in 380Vdc data loads. Starting with the motivations by looking at the development of 380 Vdc, its current implementation as a ±190-Vdc distribution with midpoint resistive grounding (shown in Figures 2.4), and the associated benefits. That led to articulating the case for LVdc as well as debunking several common myths that no longer apply. Some notable examples of 380-Vdc data centers from around the world. Figure 2.4 EPRI 380V-DC data center distribution [14] 23

42 There had been an increased attention in the field of alternative energy on the input side of the distribution side along with the development of the load side of a DC distribution system. The systems aspect of the microgrid allowed for intelligent interaction with the various national grids for things like grid load shedding and peak cost avoidance. Compared to AC distribution systems, with smart DC interconnection of DC loads such as lighting, appliances, IT equipment, and cooling systems along with PV and wind energy sources, building energy use could be optimized. For instance, according to the investigation of Lawrence Berkeley National Laboratory (LBNL) on the efficiency of power distribution in data centers, "a 28% improvement in efficiency if DC distribution is adopted." Figure 2.5 compared the two approaches, with ac presented in (a) and DC in (b) [14] Hybrid AC/DC Microgrid and Applications Because increasing people were more aware of environment problems caused by coal or gas fired generators, recently, more renewable power sources were connected in low voltage distribution systems which operated as distributed generations or as AC microgrids in order to reduce various environmental problems [15]. Conventionally, the distributed renewable sources were connected to local AC utility grid to supply local loads. In this case it was unnecessary to use long distance high voltage transmission [16]. 24

43 Figure 2.5 (a) Typical AC distribution vs. (b) Standard 380V DC distribution [14] 25

44 In an AC microgrid [17, 18], DC power from PV and fuel cell systems was converted into AC power using DC/AC inverters and wind turbine generators (WTGs) were usually connected into the AC power grid using power electronics AC/AC conversion technique. In such case, DC loads were connected to AC micro grids using AC/DC converters. Under this condition, conversion steps were necessary for DC loads in an AC microgrid with PV sources. The DC micro grids were proposed in [19, 20] in order to reduce the conversion from DC to AC. However, AC sources in a DC grid needed to be converted into DC, in the meantime AC loads were connected into DC grid using DC/AC inverters, as a result, and the efficiency was significantly reduced due to multistage reverse conversions in an AC or a DC grid. In order to minimize the energy loss resulted from reverse conversion; the concept of a hybrid ac/dc micro-grid was generated, as demonstrated in Figure 2.6. AC and DC networks in a hybrid grid were typically connected together through a four-quadrant operating three phase converter which could serve as an inverter or rectifier. AC power sources such as wind turbine generators, conventional AC loads and diesel generators were connected to the AC grid. DC sources such as photovoltaic arrayed with boost converters, fuel cell generators and DC loads were tied to the DC grid. 26

45 Voltages at the AC and DC grids were controlled, in order to satisfy the corresponding AC and DC loads requirements. The main converter was designed to operate at bidirectional power flow operation mode to incorporate complementary characteristic of wind and solar sources [21]. Utility System Static Switch (Open during Islanding Mode) Interlinking Converters Localized AC Loads AC Sub-Grid Localized DC Loads Other Loads + + DC Sub-Grid Hybrid Microgird Figure 2.6 An example of hybrid AC/DC microgrid Just like traditional AC or DC microgrid, the hybrid AC/DC microgrid can also operate in two modes. In grid-tied mode, the main converter s responsibility is to exchange energy between the AC and DC buses as well as provide some reactive 27

46 power when necessary. The converter acts as an inverter and inject power to the AC side or vice versa when the output power of the DC sources is greater than the DC loads. The hybrid micro-grid will inject power to the utility grid when the total output power of DC and AC sources is greater than the total DC and AC load. If the renewable sources cannot produce adequate power, the hybrid grid will receive energy from the utility grid. In off-grid mode, the ac voltage reference cannot be provided by the utility grid. Under this situation, battery becomes quite critical for both energy balance and voltage stability requirements through appropriate battery banks charging and discharging control. In order to demonstrate the plug and play feature of the hybrid grid, other sources and energy storages can also be added to AC and DC buses. In both operating modes, the Maximum Power Point Tracking (MPPT) control of PV and WTG systems under variable weather conditions had also been implemented in order to harness the maximum power. Figure 2.7 showed a typical hybrid system configuration which consists of AC grid on the left panel and DC grid on the right. The AC and DC grids had their corresponding sources, loads and energy storage elements, and were interconnected by a three phase converter. The AC bus was connected to the utility grid through a transformer and circuit breaker. 28

47 Ultra- Capacitor DC/AC Converter with Controller Bidirectional AC/DC Main Converter DC/AC Converter with Controller Wind Power Diesel Generator M DC/DC Converter with MPPT PV Utility Grid Breaker Charging and Discharging Converter Battery Localized AC Loads Localized DC Loads AC BUS DC BUS Figure 2.7 Typical hybrid AC/DC microgrid structure 2.2 Control Level - Droop Control in Microgrids State of the Art Control algorithm was critical for all types of microgrid. As a significant part of every power conversion architecture, control method was essential for integrating renewable energy to the microgrids [22]. Distributed sources and loads were clustered together in microgrids, providing the benefits of large number of controllable variables. Apart from this advancement in microgrids, there had been a rapid development of various essential power conditioning interfaces and their associated control algorithms for integrating multiple 29

48 micro sources to the microgrids, and then integrating the microgrids to the traditional AC power utility [22]. Given such degree of freedom, microgrid operation became highly flexible, which allowed it to operate freely in the grid-connected or islanded mode of operation [23-26]. As for the grid-connected mode of operation, only when the power grid is much larger in capacity, each micro source can be operated like a current source with maximum power transferred to the grid. In contrast, the islanded mode of operation with more stringent supply demand balancing requirements would be triggered and the control objectives would change if the mains grid was not comparatively larger or was simply disconnected resulted from the occurrence of a fault. Because there was not a strong grid and a firm system voltage, each micro source needed regulate its own terminal voltage within an allowed range based on its internally generated reference [27 30]. As a result, the micro source resembled a controlled voltage source, in which the output shared the load demand with the other sources. In order not to overstress any individual entity, the sharing should at best be proportional to their power ratings. Moreover, if the overall system operation failed, the sharing should be achieved with no or minimal communication link, thereby resulting in the minimum or even no detrimental effect to the overall system operation. 30

49 This was particularly important because most micro sources were widely dispersed making it impractical to link them by wires. Avoiding the wiring would lead to the constrain measurements being taken only within the local vicinity of each micro source. Up to now, these criteria could only be satisfied by the droop control method (both for AC and DC microgrids), in which virtual inertia was intentionally added to each micro source [27-31] Droop Control in AC Microgrids Since AC distribution dominated traditional grids, most existing droop control techniques focused mainly on AC microgrids [27 37]. Researchers had done a lot of work investigating droop control applied to an AC sub grid with at least two paralleled sources In principle, each source, employed two droop control equations for the purpose of determining its reference frequency f and voltage amplitude V a from its locally measured active P a and reactive Q a power values, respectively., the two droop equations assigned to unit x can appropriately be written as the following equation if subscript x was used for representing the source index in the AC sub grid. f f m P (2.1) * ' x x x a, x 31

50 V V n Q (2.2) * ' a, x a, x x a, x where f x and V a,x were the maximum frequency and voltage amplitude at no load, and m x and n x were the negative droop coefficients included for representing the gradual, negatively tilting gradients presented in Figure 2.8. According to (2.4), the droop coefficients should rightfully be tuned in order to make the sources share the load demand in proportion to their ratingss a,x. m S m S m S (2.3) x a, x 2 a,2 1 a,1 n S n S n S (2.4) x a, x 2 a,2 1 a,1 Once the network reached a steady state, it would have only one prevailing frequency, which was represented by the single dashed horizontal line drawn in Figure 2.8(a). At that frequency, the sum of active powers from all sources would satisfy the total demand requested by the loads. Moreover, the fractional contribution of active power produced by each source would also be in line with its rating normalized by the total capacity of the sub grid. However, this is not the case for reactive power sharing, where different line impedances between the sources and point of common coupling (PCC) commonly cause the source voltage amplitudes to be different. 32

51 Frequency (p.u.) Unit 2 Unit 1 Steady-State Points (Single Network Frequency) Real Power (p.u.) (a) Voltagte (p.u.) Unit 2 Unit 1 Steady-State Points (Different Terminal Voltage) Reactive Power (p.u.) (b) Figure 2.8 Active and reactive power droop characteristic for AC microgrid 33

52 Other system mismatches could also lead to differences, which were demonstrated in Figure 2.8(b), where two dashed horizontal lines drawn. Because of these differences, reactive power sharing became parameter dependent, and even deviated slightly from the intended. As suggested in [35] and [37], approaches aiming compensating the deviation had already been explored Droop Control in DC Microgrids Though rarely known, it is possible to droop control applied to dc microgrids In fact, this has been discussed more than a decade back in [38]. The proliferation of photovoltaic (PV) generation, fuel cells, and energy storages such as batteries and other capacitive alternatives are unable to prohibit the growing interest in DC microgrids. A DC sub grid is simpler because it has no reactive power, frequency, and phase considerations compared with an AC sub grid. The active power was represented as P d,y and voltage magnitude V d,y, where subscript y indicating the source unit number in the dc sub grid. When related by (2.5), these two quantities formed the droop equation for the dc sub grid [38] V V v P (2.5) * ' d, y d, y y a, y 34

53 Where V d,y indicated the maximum source output voltage under no load condition and v y represented the droop coefficient. When applied to more sources, (2.5) would roughly lead to proportional active power sharing if their droop coefficients were tuned as to that of (2.6), where S d,y represented the kva rating of source unit y v S v S v S (2.6) y d, y 2 d,2 1 d,1 Figure 2.9 illustrated an example of how (2.5) operated for the simple case of two dc sources. Different from an AC sub grid where a single steady-state frequency dominated, voltage magnitudes of the dc sources were usually different mainly due to finite line impedances between them and the PCC and parameter mismatches. Similar to reactive power sharing an AC sub grid, these differences caused power sharing among the sources to deviate slightly from the intended proportional distribution. Thus, techniques for improving dc source sharing could be borrowed from those for ac reactive power sharing, which presumably were more established [35-37]. 35

54 Voltagte (p.u.) Unit 2 Unit 1 Steady-State Points (Different Terminal Voltage) Real Power (p.u.) Figure 2.9 Active power droop characteristic for DC microgrid Droop Control in Hybrid AC/DC Microgrids The control algorithm for hybrid AC/DC microgrid was also a new and heated topic for researchers. However, since the extension to a hybrid spanned across at least two subgrades of either ac or dc form, it was generally more complicated Hence, the droop control in hybrid microgrid was unable to be realized by solely depending on the droop-controlled sources. Equal attention should be given to the interlinking converters, whose responsibility was to link the two types of subgrades together in a 36

55 properly managed manner. Such interlinking control was more challenging because of the following reasons: 1) The requirement of bidirectional power flow: Different from unidirectional sources, the interlinking converters needed to take care of bidirectional active power flow between the two types of sub grids, where positive and negative polarities referred to forward dc ac and reverse ac dc energy flows, respectively. 2) Double job: The interlinking converters had two roles to fulfill at every moment. On the one hand, they appeared as sources to one sub grid where energy being injected, on the other hand, they served as loads to the other sub grid where energy was taken. 3) Coordinated control scheme: Before arriving at the final active power command being transferred by the interlinking converters for proportional active power sharing, there were two different sets of droop equations for merging. Designing a coordinated droop control scheme for the purpose of controlling the hybrid microgrid so that power would be shared among the sources proportional to their power ratings instead of physical placements within the hybrid microgrid. Such sharing could be achieved solely by controlling the interlinking converters to transfer the right amount of energy between the two types of subgrids, in which the value was decided by the designed droop control scheme. 37

56 As mentioned earlier, ac distribution had been presently dominant, and would be so for many more decades. Hence the most favorable scenario would be the presence of both dc and ac subgrids with sources, storages, and loads appropriately distributed between them. The subgrids could subsequently be tied together through interlinking converters to form hybrid ac dc microgrid. Therefore, the control of hybrid microgrid would gain more attention in the near future. 2.3 Converter Level - Power Electronics Building Block (PEBB) State of the Art The Concept of PEBB As illustrated in Figure 2.10, PEBB -Power Electronics Building Block was a generic strategic ONR concept incorporating several technology aspects, which were viewed as essential to reduce cost, losses, size and weight of power electronics. By adopting building blocks that could be used for multiple applications, volume production would increase, and engineering effort, design testing, onsite installation and maintenance work for specific customer applications would reduce. Mature applications required continued costs reduction, which might be achieved at this stage through a reasonable standardization on product or system level [39]. There were two key functional components enabling a platform-based approach in power electronics. One was the Power Electronics Building Block (PEBB); the other 38

57 was the control (PEBB Power Electronics controller), which contained the control hardware and software. Power electronic systems today asked for a completed and commissioned system to validate the design cycle. Unfortunately, paper documents were unable to meet the requirement of the complexity of the next generation power electronic systems. A critical part of physics based design was validation and incremental prototyping. A new design of power electronics system that used the same PEBB as a previous design needed only to validate the new application function and design elements. Figure 2.10 Power Electronics Building Block As a broad strategic concept, Power Electronics Building Block (PEBB) incorporated progressive integration of power devices, gate drives, and other components into building blocks with defined functionality and interfaces serving 39

58 multiple applications resulting in reduced cost, losses, weight, size, and engineering effort for the application and maintenance of power electronics systems. The corner stones of the Power Electronics Building Block concept were modular and hierarchical design principles. Based on functional specifications of power electronics building blocks that related to the performance requirements of intended applications, the designing of PEBB addressed the details of the device stresses, stray inductances and switching speed, losses, thermal management, protection, measurements of required variables, control interfaces, and potential integration issues at all levels [39]. The idea of open plug-and-play architecture entailed building power systems similar to that of personal computers. The system was familiar with the PEBB capabilities, its manufacturer, and its operational requirements. In order to support the integration of these PEBBs, the overall control architecture was supposed to have the inherent capability regardless of in which way they were configured together. Each PEBB maintained its own safe operating limits. In the long run, PEBBs might be plugged into power electronics systems and operational settings automatically. Speaking of the next generations of power electronics, key technology aspects included providing economic and performance rich solutions [40]: Standard PEBB designs to cover wider market base advanced devices, integrated packaging, 40

59 progressive integration from device level to PEBB level, snubber-less design, advanced converters that provide control of reactive power, can act as active filters, minimum filter requirements, standardized control and protection architecture defined interfaces that allow "plug and play", Advanced interactive simulation tools. The single switch cell had a power switch section, a gate or base control, heat removal, and a capacitive, inductive, resistive interconnection between its neighboring cells. Hence, a building block process was fundamental to power electronic systems Plug and Play Open plug and play architecture referred to building power electronics systems in much the same way as personal computers. Power modules would be plugged into their applications and operational settings made automatically. The application was aware of what to plug into it, who made it, and how to operate with it. Each power module maintained its own safe operating limits. Realization of this vision required a community to develop standard interfaces and protocols. One purpose for plug and play architecture was to reduce cost and increase application. However, the resources to supply the demand for new power electronics 41

60 products were lacking and capable people who were proficient at it were harder to find. As a consequence, the next generation engineer wanted to replace a power electronic engineer with a computer or software engineer. Their desire was to design systems on their computers, and similar to how their PC components came together, they also hoped that power parts could come together. With the application expertise built into the equipment, open architecture asked for more designer efforts, in the meantime made his expertise available for many more applications. Moreover, the partitioning needed to implement plug and play, which allowed concentrated efforts within the partitions and the development of high volume processes for partitioned technologies. Therefore, the availability of power electronic engineers and resources were further leveraged to meet a broader range of market opportunities. Both the increased utilization of resources and high volume processes resulted in a reduction of cost. One advantage of plug and play is allowing each section of the power equipment to function independently from others. For instance, the topology controller permitted complex topologies to be transparent to the user, like soft switching, the application controller generated pulse width modulation, PWM, and drove the topology controller with the resultant signals. Soft or hard switching topologies were selected based upon performance trades. The user would drive soft switching topologies in the same 42

61 fashion as he/she did hard switching. The extra complexity and control needed for soft switching was designed into the topology controller. Similarly, power-switching blocks could be applied regardless of which type of switching device was used. The type of switching device could be selected on the basis of the best type for the job. Though taking into account the device, the power switch controller would adjust appropriately. There was no need for topology controller to be concerned with the type of the device. The topology controller managed the switch block in the same manner even if a different type of device of used. The user reaped the benefits of the best device for the job without worrying about any eccentricities specific to a particular device. [39] Modeling and Hierarchical Control of PEBB Integration and snapping elements together required intelligence and hierarchical control. Control partitions were supposed to compliment the spatial partitions or blocks. Enough intelligence and control embedded into a switch cell or two-terminal PEBB were needed to enable them to be snapped together to form higher order PEBBs. Beginning with a switch cell, in order to allow two cells to be snapped together to form a voltage source or current source phase leg, the embedded intelligence was needed. A next layer of intelligence would allow two voltage-source phase legs to form an H-bridge or three voltage source phase legs to form a three-phase bridge. 43

62 Moreover, control architecture was temporal as well as spatial. The six main sections of power converters included the power switches, gate drive, power circuit or topology manager, application or load manager, system controller and filters. Each section operated predominately in a time as well as spatial domain. The gate drive or power switch controller governed the domain no longer than 10psec. A circuit or topology controller is defined at a slower period, greater than 10psec and less than 1m sec. An application or load controller is defined from 1 ms to seconds, finally, the system controller addressed the final power electronic domain. The power switch and the gate drive sections could be taken as one through applying the requirement of Plug and Play. Knowledge on both sides of the interface was required for Plug and play. Moreover, an exchange of knowledge across the interface allowed the sections to be put together and their interrelated functions enabled automatically. In a personal computer, software drivers and BIOS programs (Basic Input Output System) were employed to effect plug and play. Microprocessors and programmable logic devices enabled controllers to exist in any number of hardware and software manifestations. The application controller could be a software program in a microprocessor and had no hardware manifestation at all. The task was to define the functions and interface requirements for each of these partitions. Not only did Protocols for information transmitted across each interface need to be defined; but BIOS and operating systems that apply to power equipment 44

63 also need to be developed. At the beginning existing computer operating systems could be used. However, power electronic necessity and sufficiency must be established. Furthermore, I/O and operating systems for power electronics need to be highly reliable and capable of real time performance. At this point, a generalized control hierarchy boiled down to the following four controllers, in which all of them were programmable and multifunctional [39]: Power switch controller Topology or circuit controller Application controller System controller The functionality of a PEBB [41, 42] as a basic building block was defined on Figure 2.11 as power conversion (single phase or multiple phases) including: power supply for gate drives & sensors stack or module assembly including gate drives voltage, current and temperature sensors including A/D conversion of sensor signals switching control incl. pulse generation for gate drive communication with control and Primary protection. 45

64 System Control Application Control Application Control -overriding control and measurement 1ms ~1s αβ to dq Transformations Id/Iq Current Control Converter Control -dq Transformation -id, iq current control 10μs ~1ms Modulator Second Level Protection Converter Switching Logic PEBB Control -Modulator -Converter Switching Logic -Second Level Protection 1μs ~10μs A/D D/A L filter Gate Drives & Protections A/D D/A PEBBs -Stack or Module Assembly -Snubbers for Safe Commutation -Gate Drive and Feedbacks -First Level Device Protection -A/D and D/A conversion -Gate Drive Power Supply -Thermal Management -AC/DC Power Terminals 0.1μs ~1μs C filter Figure 2.11 PEBB concept for power electronics 46

65 The interfaces of a PEBB were defined as following [40]: Auxiliary power interface Control interface Cooling interface Power interface 47

66 48

67 Chapter 3 PEBB Based Power Conversion Architecture for DC Microgrids 3.1 Proposed Architecture To solve this problem, a distributed power conversion architecture, which has been adopted for power supplies in communications and computing industry for years is proposed [43-46]. In those applications, distributed power conversion architectures can reduce system cost and increase efficiencies by moving from a centralized single-stage power conversion to a distributed multistage power conversion [44-46]. These concepts have not been applied or investigated in current power systems with renewable sources yet and they will have a great potential for microgrids or other electrical energy systems. Shown in Fig. 1.4, the proposed architecture is applied in a microgrid and different renewable energy source and loads can be integrated into it as many as possible [47-49]. 49

68 Figure 3.1 Proposed distributed power conversion architecture 50

69 Battery banks can be deployed locally near to wind generators, solar farms and micro turbines. Functions of centralized converters have been assigned and distributed to different stages. A stage is a collection of similar sub functions and connected to other stages by DC buses of different voltage levels. Power electronics converters in the same stage perform similar functions but much less and simpler compared to centralize converters. For instance, converters in the first stage reduce mismatches among solar modules and battery cells, draw maximum power from these sources and deliver it to the next stage. Voltage transformation and instantaneous energy storage are realized in the second stage. Power converters in the third stage provide real and reactive ac power to the grids, enhance power quality and reliability, support grid voltages, correct voltage sags, and provide uninterruptible power supply functions to the loads when the grids are faulty [47]. Communications among these converters are unnecessary for basic operations. But there will be a centralized controller called Energy Manager to coordinate the whole operation [47]. Compared to the conventional centralized design, this architecture has the following advantages: 1. Power electronics converters in the same stage only need to perform a few simple tasks. Their performance can be optimized independently and won t degrade as tasks increases in the future. 51

70 2. Mismatches among solar modules or battery cells can be significantly reduced. The output capacity of renewable sources will be increased. 3. The total power extracted from renewable energy sources can be boosted with individual maximum power point tracking (MPPT). 4. Reliability is improved because functions of each converter have been simplified. Meanwhile, because converter modules are in parallel, failures of a few of them will not shut down the whole system. 5. Converters can be modularized. Manufacturing cost will be reduced by massive production. 6. DC buses acts as a buffer among individual renewable sources, batteries and micro-grid, so ac buses (output) will experience a smoother power flow. 7. The proposed distributed architecture can enhance the reliability of power delivery and also increase the penetration of renewable energy into the grid. As the complexity of each dc-dc converters has been reduced, high efficiency resonant converters can be adopted in this architecture, and the total efficiency could be higher than AC micro grids [44]. 8. This architecture can replace heavy and costly line frequency transformers with lightweight and less expensive high frequency transformers. Extra cost with the installation and site preparation can be saved. 52

71 Among various options for the distributed power conversion architecture, the DAB uniform design represents an inexpensive means for converting energy from renewable sources into electrical energy for interface to the grid. The primary drawback of the DAB uniform design is high initial cost. 3.2 Realization in UW-Milwaukee Microgrid Using the techniques and methodologies from the previous chapters, a 12 bus system model is created for the University of Wisconsin - Milwaukee campus. Figure 3.2 shows the physical layout of the system. Each number represents a bus. Bus 1 represents the collection point for multiple offshore wind turbines located in Lake Michigan. It provides around 3 MW. Figure 3.2 University of Wisconsin-Milwaukee microgrid map 53

72 In order to complete the model, system parameters calculated from the physical dimensions are needed. The areas for the buildings are estimated from satellite images; the solar capacity and load estimate are created based on the square footage and techniques described in previous sections, which is shown in Table 3.1. Table 3.1 Summary of loading and solar capacity per building Name Area (ft 2 ) Stories Total Area (ft 2 ) Load (kw) Solar(kW) Bus # Physics Building Physics Building EMS EMS small Knuckle Center UWM Union Vogel and Curtin Mitchell Hall Mel encamp Hall Art Building Pearse and Garland Hall Johnston and Holton Hall Sabin Hall

73 Klatches Pavilion Library Ender Hall Norris Health Center Chapman Hall Architecture Building Engleman and Cuningham Hall Lapham Hall Lubar Hall Chemistry Building Total Table 3.2 shows the solar capacity and load per bus. The resistance is calculated using load in kw and the bus voltage of 380 V. The capacitance is calculated using the number of converters and an individual converter capacitance of 1 mf. 55

74 Table 3.2 Bus summary of loading and solar capacity Bus Bus Load (kw) R load (ohms) Generation (kw) # of Converters C (F) 1 Line Parasitic

75 3.3 DC Bus Voltage Level DC Bus Voltage Identification When the physical layout has been determined, the most important properties of the system shift to the voltage level. It is difficult to select the appropriate system voltage because of many competing factors, such as efficiency, safety, and various standards. Figure 3.3 illustrates various voltage levels, standards, implementations, and benefits. The main voltage constraint in the US is the NEC s 600 V, the maximum for low voltage for residential areas. It is important to note that equipment becomes much more expensive when the voltage rating goes above 600V. Figure 3.3 DC bus voltage selection overview 57

76 The380 volts are highlighted in Figure 3.3 because it provides good efficiency though it is below the 600 volt maximum. Additionally, there is an ETSI Standard in progress for the 380 VDC voltage level (EN ). Several telecommunication facilities have also adopted this voltage level. Figure 3.4 shows the estimated energy usage for various voltages. The 400 VDC systems use less than 90% of the energy in a traditional AC system. Based upon all of the benefits mentioned above, the system voltage level is selected as 380 V. Figure 3.4 Comparison of energy usage for different system voltages 58

77 3.3.2 The Advantages of 380V DC Bus Voltage Subsequent R&D is able to fully articulate the benefits and advantages of 380 Vdc. They show that 380 Vdc: 28% more efficient than 208 Vac systems 15% less up-front capital cost in production volumes 33% less floor space 36% lower lifetime cost times less copper than -48V DC systems Higher Reliability: Higher reliability is the most outstanding advantage of 380 Vdc. By eliminating the unnecessary conversion architectures, such as the PDUs and transformers, the front end of the PSU, and in particular, the inverter that is on the output side of the UPS in the ac design, the reliability of the power delivery chain is improved by 1000%. This should not come as a surprise because telecommunications companies have been using -48Vdc systems for decades. In fact, in 2010 when Intel surveyed its manufacturing organization, it was found that a plenty of critical failures that year was the failures in the data centers that supported manufacturing [14]. Additionally, the vast majority of these failures were resulted from errors with manufacturing organization. Higher Efficiency: 59

78 380 Vdc is 28% more efficient than the current practice of 208 Vac. The Green Grid did a peer review of the LBNL findings and concluded that the efficiency findings were real. The report especially emphasized that if certain techniques be adopted by ac distribution on 380 Vdc, then the efficiency could be improved by 5 7%. In the Intel, HP/EY P, and Emerson Network Power study, the improvement of efficiency was about 7% for dual-corded installations and 8% otherwise compared to 415 Vac. Real implementations at Duke Energy and Green.ch found 15% and 10% efficiency improvements, respectively. So an existing legacy data center power distribution system is replaced, the efficiency gain could be as high as 28%. The gain could be achieved at least 7 8% for a Greenfield data center, depending on whether it uses a dual-corded design or not [14]. Lower Cost: Given that the power distribution system and eliminating components have been vastly simplified, the cost in volume production is expected to be 15% less. ABB and Green.ch reported that the cost for the 380-Vdc system built in Zurich in 2010 was 10% less than for the ac system. (The Green.ch data center has both.) In fact, lower capital and operating cost are the reasons explaining why 380 Vdc is the ideal voltage. Another important component in making a cost-effective, efficient voltage distribution standard is to stay below 420 V so that other parts can share the volume economics with desktop personal computer PSU s. The final reason of why the 60

79 industry selected 380 Vdc is because of a specification that requires operation up to 400 Vdc and successful survival when exposure to up to 410 Vdc) [14]. Less Space and Fewer Materials: The Intel, HP/EY P, and Emerson Network Power study found that the footprint of the dc power distribution system was 33% smaller than for ac. The power supply was also reduced 30% by volume due to the elimination of components. Although this is unlikely to make PSU s smaller since the servers will desire to use either ac or dc PSU s during the changeover, it will make the 380-Vdc PSU s more sustainable. They use less of the earth s resources, and there is less to recycle at the end of their lives. With respect to the wiring plant, 380 Vdc uses less copper than ac, even 415 Vac. Figure 3.5 shows that at the same power conducted, relative conductor diameter for 380V DC uses less copper than even the high-efficiency 415V AC. Conductors for 380V DC are much smaller than -48V DC with much less weight and less copper cost. Because of skin effect and the reactive current overhead, ac conductors need to be sized bigger than dc for the same voltage and power capacity. We can see why telecommunications companies and the island nation of Japan are looking to replace 48Vdc with 380 Vdc across the board. In fact, some organizations are thinking about paying for the conversion simply because of the salvage value of the 48Vdc copper bus bars and conductors that will be replaced. 61

80 Figure 3.5 Comparison of conductor diameters (courtesy of UECorp) [14] 3.4 PEBB Topology Selection Bidirectional Dual Active Bridge The overview of PEBB Topology Candidates In order to respond to the increasing demand for an intermediate storage of electrical energy in battery systems as well as the request for using renewable energy, PEBBs have developed to be able to handle bidirectional power flow and in the same time provide galvanic isolation. Battery charging systems, photovoltaic equipment and 62

81 auxiliary power supplies in proposed DC microgrid applications are quintessential examples that have been created to address these concerns. Just as mentioned in chapter 2, qualified DC/DC converter candidates of PEBB should be able to fulfill the requirements of the following general specifications: Galvanic isolation Bidirectional power flow ability High conversion efficiency in the target operating range Low converter volume Constant switching frequency More detailed specifications for the proposed power conversion architecture are listed in Table 3.3 Table 3.3 Detailed specifications of power converters Converter Description Input Voltage Output Voltage Power (kw) DC-DC PV 48 VDC 380 VDC (±190 VDC) 2kW DC-DC Battery 48 VDC 380 VDC (±190 VDC) 2kW(Bidirectional) AC-DC Wind Based on Power 380 VDC (±190 VDC) 2kW DC-DC Load Step Down 380 VDC (±190 VDC) 48 VDC 2kW 63

82 DC-AC Grid Connection 380 VDC (±190 VDC) 480 VAC 7.5MW(Total) From what has shown in Table 3.3, it is not difficult to differentiate the family of isolated bidirectional DC/DC converter from others having the potential to realize PEBB concept in the proposed power conversion architecture. The different components of a bidirectional DC/DC converter with galvanic isolation are depicted in Figure 3.6. V in + - Filter Network DC/AC Resonant High Frequency Network High Frequency Transformer Resonant High Frequency Network AC/DC Filter Network + - V out High Voltage Side Low Voltage Side Figure 3.6 General structure of an isolated, bidirectional DC/DC converter Different blocks in the diagram represent different components required for an isolated bidirectional DC/DC converter [50]: Filter networks provide smooth terminal voltages and current. DC/AC converter is a switch network which provides AC power to the HF transformer and the AC/DC converter supplies DC power to the receiving port; both converters allow for bidirectional power transfer. 64

83 The reactive HF networks are used to modify the shapes of the switch current waveforms in order to achieve low switching losses. The HF transformer is required in order to achieve electric isolation; it further enables large voltage and current transfer ratios.(small volume at HF) Bidirectional DC/DC converter topologies with a system configuration based on Figure 3.6, are called Single-Stage Topologies because they contain a minimum number of conversion stages. Accordingly, the total number of required components is comparably low. However, the operation within wide input and output voltage ranges causes ineffective transformer and switched utilization. Improved transformer and switch utilization will be achieved with multi-stage topologies, which contain an additional power converter in order to adjust voltage and current levels [50]. Here, the single-stage topologies are discussed first, and the issue of multi-stage topologies will be stressed in the following section. The family tree of the single-stage and two-stage bidirectional and isolated DC-DC converter topologies are illustrated in Figure

84 Bidirectional, two port DC/DC converters with galvanic isolation Single-Stage Topologies Two-Stage Topologies Basic Single-Stage Converters Dual Bridge Converters without Resonant Network Dual Bridge Converters without Resonant Network Isolated and Bidirectional Flyback Two Voltage Sourced Ports Voltage and Current Sourced Ports Bidirectional seriesparallel resonant LLC converter Isolated and Bidirectional Forward Single Phase Dual Active Bridge Voltage sourced full bridge with a current sourced full bridge Bidirectional seriesparallel resonant LCC converter Isolated and Bidirectional Ćuk Three Phase Dual Active Bridge Voltage sourced full bridge with a current doubler Voltage sourced full bridge with a pushpull structure Figure 3.7 Family tree of single-stage and two-stage bidirectional isolated DC/DC topologies 66