An Active Method for Implementing the Unintentional Islanding Test in Distributed Generation Systems by Michel E. AlSharidah B.Sc., The University of Arizona, 1995 M.Sc., Portland State University, 1999 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in The Faculty of Graduate Studies (Electrical and Computer Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) December 2012 © Michel E. AlSharidah 2012 Abstract A development of an island stabilizing element (ISE) for use in the IEEE 1547 unintentional islanding test is introduced. The new test setup for nonislanding inverters interconnected with the grid is proposed. The current testing standard uses discrete RLC elements to simulate the test-island. Even though the RLC simulated test-island is useful for its reproducibility, relative scalability and short setup time, as inverter power ratings increase so does the size and cost of the RLC simulated island. The proposed island stabilizing element can represent the function of the resonant part of the test island as well as provide compensation for dynamic changes in power during the test for producing near worst case conditions for an islanding test. This work introduces improvements to the unintentional islanding test. The island stabilizing element is designed and developed. Test cases proved the efficient application of the ISE as means to replace the LC elements in the unintentional islanding test. ii Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Island Definitions . . . . . . . . . . . . . . . . . . . 1.2 Islanding Prevention . . . . . . . . . . . . . . . . . . 1.2.1 Passive Methods . . . . . . . . . . . . . . . . 1.2.2 Active Methods . . . . . . . . . . . . . . . . 1.2.3 Alternative Methods . . . . . . . . . . . . . 1.3 Standard Test of Unintentional Islanding . . . . . . 1.4 Advantages and Shortcomings of the Standard Test 1.5 Motivation and Dissertation Outline . . . . . . . . . 2 The 2.1 2.2 2.3 2.4 2.5 Standard Unintentional Islanding Test . . Simulation Environment . . . . . . . . . . . . . Test Circuit . . . . . . . . . . . . . . . . . . . . Simulated Island Parameters . . . . . . . . . . . Test Procedure . . . . . . . . . . . . . . . . . . . Simulation Results of the Standard Unintentional 2.5.1 Equipment Under Test (EUT) . . . . . . 2.5.2 Test Start/Stop . . . . . . . . . . . . . . 2.5.3 Case 1: The Single-Phase Islanding EUT 2.5.4 Case 2: The Single-Phase EUT with AFD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 2 2 5 10 11 13 14 . . . . . . . . . . . . . . . Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 18 19 20 21 21 21 22 22 27 . . . . iii Table of Contents 2.5.5 2.5.6 2.5.7 Case 3: The Single-Phase EUT with RPV . . . . . . Case 4: An Islanding Three-Phase EUT . . . . . . . . Case 5: A Three-Phase EUT with Negative Sequence Injection . . . . . . . . . . . . . . . . . . . . . . . . . 30 31 3 Proposed Island Stabilizing Element (ISE) . . . . . . . . . . 3.1 Overview of Advantages and Disadvantages of Standard Islanding Test . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Modified Unintentional Islanding Test . . . . . . . . . . . . . 3.3 Island Load P & Q . . . . . . . . . . . . . . . . . . . . . . . 3.4 ISE Representing Qload . . . . . . . . . . . . . . . . . . . . . 3.5 ISE Representing Pload & Qload . . . . . . . . . . . . . . . . . 3.6 ISE Representing Mismatch in P and Q . . . . . . . . . . . . 3.7 Initial Design of the ISE . . . . . . . . . . . . . . . . . . . . 3.7.1 Small-Signal Model of the Single-Phase ISE . . . . . 3.7.2 PI-Controller Design . . . . . . . . . . . . . . . . . . 3.7.3 PI-Controller Analysis . . . . . . . . . . . . . . . . . 3.7.4 Average and Ripple Current . . . . . . . . . . . . . . 36 4 Improved Control Design for the ISE . . . . . . . . . . . . 4.1 Synchronous Frame Model of the Single-Phase ISE . . . . . 4.2 Synchronous Frame Model of the Three-Phase ISE . . . . . 4.3 Proposed Synchronous Frame Digital Current Control . . . 4.3.1 Current Reference Calculation . . . . . . . . . . . . 4.3.2 Synchronous Frame Current Control . . . . . . . . . 4.4 Space Vector PWM (SVPWM) Generation . . . . . . . . . 4.5 Verification of Islanding PWM Inverter Performance . . . . 4.5.1 Verification of Single-Phase PWM Inverter Islanding 4.5.2 Verification of Three-Phase PWM Inverter Islanding 4.5.3 Islanding Detection Method Employed . . . . . . . . . 49 49 51 54 54 55 60 64 64 66 66 5 Modified Unintentional Islanding Test . . 5.1 Introduction . . . . . . . . . . . . . . . . . 5.2 Case 1: A Single-Phase Islanding Inverter . 5.2.1 Inverter Voltage & Frequency . . . 5.2.2 Inverter Real and Reactive Power . 5.3 Case 2: A Single-Phase Inverter with AFD 5.4 Case 3: A Single-Phase Inverter with RPV . . . . . . . 71 71 72 73 74 75 78 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 36 36 37 38 39 40 40 42 43 44 46 iv Table of Contents 6 Experimental Development . . . . . . . . . . . . . . 6.1 Hardware Circuit . . . . . . . . . . . . . . . . . . . 6.2 Simulated Inductance (Lsim ) . . . . . . . . . . . . . 6.3 Simulated Capacitance (Csim ) . . . . . . . . . . . . 6.4 ISE Series Inductance and Switching Frequency . . 6.4.1 ISE Current Ripple Experimental Results and sis . . . . . . . . . . . . . . . . . . . . . . . . 6.4.2 Lsim ISE Current Ripple . . . . . . . . . . . 6.4.3 Csim ISE Current Ripple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analy. . . . . . . . . . . . . . . 83 83 83 84 84 7 Conclusion . . . . . . . . . . . . . . . . . 7.1 Improvements to the Standard Test . 7.2 Limitations of Proposed Test Setup . 7.3 Broader Application and Future Work . . . . . . . . 92 92 93 93 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 85 85 Appendices A Space Vector PWM . . . . . . . . . . . . . . . . . . . . . . . . A.1 Three-Phase PWM Inverter Model . . . . . . . . . . . . . . . A.2 Sampled Space Vector Phase Angle . . . . . . . . . . . . . . A.2.1 Solution for t1 and t2 in Sector S1 . . . . . . . . . . . A.2.2 Solving for t2 and t3 in Sector S2 . . . . . . . . . . . A.3 Sampled Phase Voltage Amplitudes . . . . . . . . . . . . . . A.4 Space Vector PWM ON-Time Durations for Three-Phase Inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 104 105 105 106 109 B Hardware B.0.1 B.0.2 B.0.3 B.0.4 B.0.5 Implementation . . . . . Inverter Using IGBT Model Driver Circuit . . . . . . . Mode Selection . . . . . . Dead Time . . . . . . . . . Voltage Measuring Circuit C Circuit and PCB . . . . . . . . C.1 Schematics of Control Board C.2 Schematics of Power Board . C.3 PCB Layouts . . . . . . . . . . . . . . . . . . . . . 115 . . . . . . . . . . . MUBW 20-06 A7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 118 119 119 119 120 . . . . . . . . . . . . . . . . . . . . 130 131 135 137 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Table of Contents D Schematics for 3-Phase Islanding Test . . . . . . . . . . . . . 143 E Unintentional islanding test conditions . . . . . . . . . . . . 145 vi List of Tables 2.1 Unintentional islanding test parameters . . . . . . . . . . . . 20 4.1 4.2 4.3 PI controller design parameters . . . . . . . . . . . . . . . . . Space vectors and corresponding switching states . . . . . . . ON-time durations for three-phase SVPWM control . . . . . 60 62 64 5.1 ISE case 1 parameters . . . . . . . . . . . . . . . . . . . . . . 72 6.1 ISE inductive or capacitive load specification . . . . . . . . . 84 A.1 ON-time durations for space vector PWM . . . . . . . . . . . 110 A.2 ON-time duration using sampled phase voltages . . . . . . . . 116 vii List of Figures 1.1 1.2 1.3 1.4 1.5 Power flow before (a) and after (b) grid-disconnection. . . . Voltage and current output of EUT implementing AFD . . Initial islanding test circuit with matched load . . . . . . . Islanding test circuit according to the German proposal [15] IEEE Std 1547 unintentional anti-islanding test circuit . . . 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 2.11 2.12 2.13 2.14 2.15 2.16 2.17 2.18 Unintentional islanding test per-phase circuit . . . . . . . . . Per-phase unintentional islanding test circuit. . . . . . . . . . Voltage and frequency . . . . . . . . . . . . . . . . . . . . . . EUT, EPS, and load currents (pu). . . . . . . . . . . . . . . . Real and reactive power . . . . . . . . . . . . . . . . . . . . . Frequency response of EUT during the standard islanding test. EUT PI-controller with AFD control scheme. . . . . . . . . . EUT current with AFD implementation . . . . . . . . . . . . Frequency response for EUT under standard isalnding test . . EUT trip time for AFD chopping fraction of 0.05 . . . . . . . EUT with modified AFD trip times (CF=±5%). . . . . . . . RPV current control loop . . . . . . . . . . . . . . . . . . . . EUT with RPV islanding test . . . . . . . . . . . . . . . . . . EUT trip times . . . . . . . . . . . . . . . . . . . . . . . . . . Real and reactive power of the inverter, load and grid . . . . PCC voltage magnitude and frequency . . . . . . . . . . . . . Inverter V and I before and after grid-disc. . . . . . . . . . . Negative sequence voltage component at PCC . . . . . . . . . 16 19 23 24 25 26 27 28 29 29 30 31 32 32 33 33 34 35 3.1 3.2 3.3 3.4 3.5 3.6 3.7 Standard unintentional islanding test . . Modified island load with R and ISE . . Modified island load with ISE only . . . Addition of ISE to island load . . . . . . Proposed test island stabilizing element Per-phase circuit diagram . . . . . . . . Closed-loop PI-Control system . . . . . 37 38 39 40 41 41 43 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 9 12 12 13 viii List of Figures 3.8 3.9 3.10 3.11 3.12 Generation of the pulse width modulated signal . . Generation of the pulse width modulated signal . . Generation of the pulse width modulated signal . . Bode plot of control to output TF . . . . . . . . . Simulated impedance of ISE relationship to current . . . . . . . . . . . . . . . . ripple . . . . . 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.19 4.20 Per-phase circuit diagram . . . . . . . . . . . . . . . . . . . . Transfer function block diagram . . . . . . . . . . . . . . . . . Per-phase equivalent circuit of a three-phase PWM inverter . Proposed synchronous frame digital current control . . . . . . Closed loop current current control in synchronous frame . . Simplified reference frame current control transfer function . Current control discrete-time block diagram . . . . . . . . . . Current control open-loop poles and zeros and desired pole z ∗ Design verification of the synchronous frame current controller Three-phase two-level inverter . . . . . . . . . . . . . . . . . . Space vector representation for all possible switching states . Switching commands generated for SVPWM . . . . . . . . . SVPWM phase A voltage outputs . . . . . . . . . . . . . . . Single-phase inverter output under islanding condition . . . . Three-phase inverter output under islanding condition . . . . AFD implemented in the single-phase PWM control . . . . . AFD implemented in the single-phase PWM control . . . . . Negative sequence voltage injection at 0, 2 and 5% . . . . . . 0%, 2% and 5% negative sequence injection . . . . . . . . . . Output of negative sequence islanding detection . . . . . . . . 49 51 52 54 55 56 56 58 59 61 61 63 63 65 67 68 68 69 70 70 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 Modified island load with ISE . . . . . . . . . . . . . . . . VPCC and IEU T under zero mismatch condition . . . . . f , V, and Iinv results . . . . . . . . . . . . . . . . . . . . . P and Q of inverter results under islanding condition . . . Voltage and frequency response of inverter with AFD . . . Real and reactive power of the inverter with AFD . . . . f comparison between the standard and modified test . . f comparison with reactive power compensation . . . . . Voltage and frequency response of inverter with RPV . . . Real and reactive power of the inverter with RPV . . . . RPV frequency results comparison . . . . . . . . . . . . . Standard and modified test frequency results comparison 72 73 74 75 76 77 78 79 80 81 81 82 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 44 45 45 48 ix List of Figures 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 6.10 6.11 6.12 6.13 6.14 6.15 6.16 6.17 6.18 6.19 6.20 6.21 Experimental system . . . . . . . . . . . . . . Control circuit board . . . . . . . . . . . . . . Power circuit board . . . . . . . . . . . . . . . The whole circuit . . . . . . . . . . . . . . . . ISE hardware prototype system . . . . . . . . Lsim = 0.5p.u. at fpwm = 20kHz . . . . . . . Lsim = 1p.u. at fpwm = 20kHz . . . . . . . . Csim = 0.5p.u. at fpwm = 20kHz . . . . . . . Csim = 1p.u. at fpwm = 20kHz . . . . . . . . L = 2.6mH, Rsim = 2p.u. and fpwm = 20kHz Current Ripple at L = 2.6mH . . . . . . . . . L = 26mH, Rsim = 2p.u. and fpwm = 20kHz Current ripple at L = 26mH . . . . . . . . . L = 2.6mH ,Lsim = 1p.u. and fpwm = 20kHz L = 26mH ,Lsim = 1p.u. and fpwm = 20kHz Current ripple at L = 2.6mH . . . . . . . . . Current ripple at L = 26mH . . . . . . . . . L = 2.6mH ,Csim = 1p.u. and fpwm = 20kHz L = 26mH ,Csim = 1p.u. and fpwm = 20kHz Current ripple at L = 2.6mH . . . . . . . . . Current ripple at L = 26mH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 86 86 86 86 87 87 88 88 89 89 89 89 90 90 90 90 91 91 91 91 A.1 A.2 A.3 A.4 A.5 A.6 A.7 Three-phase two-leve inverter . . . Sectors of SVPWM . . . . . . . . . Sector 2 . . . . . . . . . . . . . . . Three-phase VSI . . . . . . . . . . (V⃗α ,V⃗β ) transformation in sector S1 Sector 2 α,β transformation . . . . Three-phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 105 107 111 112 114 116 B.1 B.2 B.3 B.4 B.5 B.6 B.7 B.8 B.9 B.10 ISE hardware layout . . . . . . . . IGBT Module . . . . . . . . . . . . Block diagram of IGBT module . . The circuit of the voltage sensor . Voltage scaling circuit . . . . . . . Voltage scaling from sensor output Voltage buffer circuit . . . . . . . . Current sensor circuit . . . . . . . Measured current scaling . . . . . . Temperature sensor circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 118 121 122 123 124 125 126 127 128 x List of Figures B.11 Temperature hysteresis loop . . . . . . . . . . . . . . . . . . . 129 C.1 C.2 C.3 C.4 C.5 C.6 C.7 C.8 C.9 C.10 C.11 C.12 PowerSupply AD.SCHDOC [38] . . . . VoltageSensor.SCHDOC [38] . . . . . CurrentSensor.SCHDOC [38] . . . . . TMPSensor.SCHDOC [38] . . . . . . . Driver.SCHDOC [38] . . . . . . . . . . MUBW20.SCHDOC [38] . . . . . . . . Control board PCB [38] . . . . . . . . Control board PCB top layer [38] . . . Control board PCB bottom layer [38] Power board PCB [38] . . . . . . . . . Power board PCB top layer [38] . . . . Power board PCB bottom layer [38] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 132 133 134 135 136 137 138 139 140 141 142 D.1 Schematics for 3-phase islanding test [38] . . . . . . . . . . . 144 xi Glossary DG Distributed Generation PV Photovoltaic EPS Electrical Power System EUT Equipment Under Test AFD Active Frequency Drift DR Distributed Resources PJD Phase Jump Detection PCC Point of Common Coupling IEEE Institute of Electrical and Electronics Engineers PSIM PowerSIM simulation software MATLAB MathWorks simulation software PWM Pulse-Width Modulation RPV Reactive Power Variation PLL Phase-Lock Loop THD Total Harmonic Distortion NDZ Non-Detection Zone SMS Sliding Mode frequency Shift MSD Main Monitoring with Allocated Switching Devices PLCC Power Line Carrier Communication SCADA Supervisory Control and Data Acquisition xii Acknowledgements I would like to acknowledge the patience and support of my advisor Professor William G. Dunford. Without him I would not have the guidance and drive to move forward. I would also like to acknowledge my colleague Wei Feng for our collaboration in this research and the long hours spent working on the hardware. I would also like to acknowledge all my colleagues in the power lab for their support. To my parents for being patient and their prayers. To Chris Westcott and his family for their constant love and support. To my late brother who was my example of living life fully and believing in me. I miss you very much! xiii Chapter 1 Introduction The popularity of distributed generation (DG) systems has improved in recent years due partially to development in performance and largely to government deregulation. For example, the increase in the number of photovoltaic PV systems naturally drives cost down as more units are produced. Generally, PV systems generate distribution level voltage and can be integrated to the utility grid using inverters. Benefits of using PV systems are not limited to only the customer. Distribution and transmission relief, peak shaving, deferral of high cost transmission and distribution upgrades are some utility benefits. From the customer prospective, effective use of energy, better power quality and reliability, tax and subsidization incentives, and positive environmental impact are some benefits customers would enjoy. In spite of these benefits there are hazards when PV systems in particular, or DG systems in general, are connected to the utility grid. When local power generation energizes a portion of the distribution network without supervision and control of the utility, it would be a source for potential hazards that can be summarized as causing: 1. harm to maintenance personnel when servicing the energized feeder. 2. damage to utility customer equipment due to voltage and or frequency being uncontrolled. 3. damage to switching or measuring devices due to unsynchronized reclosure 4. malfunction of automatic re-closing devices. The previous concerns are amongst other factors that highlight the importance of the standard of interconnection of distributed resources (DR) with the utility. IEEE standard 1547-2005 defines islanding and sets a test to measure the effectiveness of islanding detection schemes. In the process of investigating the standard’s test of unintentional islanding, a clear definition of an island becomes critical for proper identification. 1 1.1. Island Definitions 1.1 Island Definitions The term island might be misleading but in power systems an island is defined as A portion of an Area EPS is energized solely by one or more Local EPSs through the associated PCCs while that portion of the Area EPS is electrically separated from the rest of the Area EPS [1]. where Area EPS is the facility of the electrical power system that delivers power to local electrical power systems (Local EPS) at a point in the grid. That point, the point of common coupling (PCC), the Local EPSs such as distributed resources (DR) could sustain the local load demand creating an island that the Area EPS has no control over. Identification of such an island is critical at the PCC for safety reasons. When islanding conditions exist, isolation of the island is critical and usually occurs on the distribution line, but islanding may also occur on transmission lines when large numbers of grid-connected inverters or other distributed generation are present. The islanding condition covered in this document occurs when the low voltage distribution lines are interrupted. A worst case scenario for this condition is when the island is localized and not including the main transformer. Another definition for an island is described as a portion of the distribution system that is self energized intentionally or accidentally and isolated from the utility grid. 1.2 Islanding Prevention Numerous islanding prevention methods have been developed. These methods can be organized in three main categories. Passive, active, and alternative methods. A review of islanding detection methods is included to better understand the worst case scenario under unintentional islanding conditions. The review will briefly describe the most common methods and some recently developed ones based on principle of detection, strengths ,and weaknesses. 1.2.1 Passive Methods Passive methods monitor selected system parameters. If monitored parameters are out of the permitted operational range, a potential island is identified 2 1.2. Islanding Prevention (a) (b) Figure 1.1: Power flow before (a) and after (b) grid-disconnection. and a cessation of energy command is issued. An advantage of implementing a passive method is its zero influence on the Area EPS. Over/Under Voltage and Frequency Detection The principle of over/under voltage and over/under frequency detection methods is to monitor the grid-connected DG inverter voltage and frequency. If either is out of the recommended threshold range1 , a potential islanding condition is confirmed and the DG inverter ceases to energize the Local EPS. The recommended threshold voltage and frequency ranges are according to [1] and [2]. At the moment of disconnection, if local generation and Local EPS load are not under balanced power conditions, there is real and reactive power mismatch supplied by the Area EPS and is defined as illustrated in Figure 1.1. The real and reactive power mismatch is the difference between the real and reactive power supplied by the DG inverter and the real and reactive power consumed by the local load, respectively, just before the moment of utility disconnection. Pload = Pinv + ∆P Qload = Qinv + ∆Q (1.1) The average inverter real and reactive power supplied is defined as Pinv = Vinv Iinv cos(ϕ) Qinv = Vinv Iinv sin(ϕ) 1 (1.2) 0 .88 ≤ V ≤ 1.1p.u. , 5 9.3 ≤ f ≤ 60.5Hz 3 1.2. Islanding Prevention Vinv and Iinv are the r ms values at the PV inverter terminals and cos(ϕ) is the displacement power factor (d .p.f ). The island load is modelled as parallel RLC components. Real power is expressed as, 2 Vinv Pload = (1.3) R reactive power is expressed as, ( Qload = 2 Vinv 1 − ωC ωL ) (1.4) After disconnection, the voltage and frequency are uncontrolled and the new operating point should satisfy the balance of real and reactive power between the DG inverter and island load. Under such conditions, the following could be concluded: • In most cases real and reactive power mismatch is large. ∆P > ±20% or ∆Q > ±5%. • If reactive power mismatch is positive (∆Q > 0), then the inverter frequency will increase until the reactive power supplied by the local EPS capacitance equals that consumed by the island load inductance. On the other hand, if reactive power mismatch is negative (∆Q < 0), then the inverter frequency will decrease until the reactive power supplied by the local EPS capacitance equals that consumed by the island load inductance. • If real power mismatch is positive (∆P > 0), then the inverter voltage will be higher than the Area EPS voltage and if real power mismatch is negative (∆P < 0), then the inverter voltage will be lower than the Area EPS voltage. Literature Opinion suggest that islanding conditions of ∆P and ∆Q falling into the none-detection zone (NDZ) of the over-voltage/under-voltage or over-frequency/under-frequency protection could be significant [3]. Phase Jump Detection (PJD) Phase jump detection (PJD) is to monitor the phase difference between the Area EPS voltage and DG inverter current. In the presence of Area EPS, the voltage source can be assumed stiff at system voltage and frequency. The 4 1.2. Islanding Prevention DG inverter operates as a power conditioner, regulating sinusoidal current impressed onto Area EPS voltage. The phase of the DG inverter current is synchronized to the Area EPS voltage via phase-lock loop (PLL). At the instant of island creation (grid-disconnection), power mismatch between island load and DG inverter will force the voltage to phase jump to a new operating condition to balance power Equation 1.1. An island is confirmed if the phase error due to the island inductance exceeds a predetermined threshold phase angle [4]. This method is simple and easy to implement using phase-lock loop by analog circuit or digital signal processor (DSP). The size of NDZ is reduced by setting the threshold phase angle ϕth to a small value. The disadvantage is with smaller ϕth , false detection can occur due to start-up of induction motor or switching power factor correction capacitor. This method also fails when power factor is near unity. Voltage Harmonic Detection The principle of voltage harmonic islanding detection is to monitor the total harmonic distortion (THD) of the DG inverter for certain change in harmonic distortion. At the point of common coupling, voltage harmonic distortion content increases significantly due to interaction between island loads (high impedance), DG inverter current harmonics, presence of nonlinear power electronic loads in the island, non-linear excitation current of distribution power transformer and other local EPS harmonic sources. In this method, islanding is confirmed when total harmonic distortion exceeds certain set threshold values [5]. This method is effective due to the fact that it does not rely on power mismatch. However, this method is computationally expensive compared to other passive methods. Also, if an island has a load with a high load quality √ factor, Qf = R C/L, it would serve as a low-pass filter for wide range of frequencies. Thus, THD might remain within threshold and detection fails. Finally, setting a value for threshold THD might be difficult. If the selected value for THD is too low 2 , it will render the method impractical due to failure in detection. 1.2.2 Active Methods The principle of this category of methods is to slightly perturb a system variable such as voltage or frequency and simultaneously observe their im2 (T HD < 0.5%) 5 1.2. Islanding Prevention pact. Islanding is detected if observed variables are forced out of threshold range. Output Real or Reactive Power Variation The principle of output power, real or reactive, variation is to perturb the output power of the DG inverter and simultaneously monitor the voltage magnitude if real power is varied, or frequency if reactive power is varied at the inverter terminals. Islanding is confirmed when the inverter voltage or frequency falls out of the threshold limits 3 . Let real and reactive power be defined as P = P̄ + ∆P (1.5) Q = Q̄ + ∆Q (1.6) where P̄ and Q̄ are the average real and reactive power, respectively. While ∆P and ∆Q are the perturbed real and reactive power, respectively. Voltage variation, ∆V , can be expressed as a function of real power variation and load real power (see Equation 1.7) [6–8]. Frequency variation, ∆ω, can also be expressed as a function of reactive power. √ ∆P R ∆V = (1.7) 2 Pload This method is robust and efficient in reducing the NDZ to zero in the case of single inverter connected to the utility network. The efficiency of this method starts to decline as more inverters are connected to the utility network. Statistically, the NDZ will increase as more inverters start varying power without any synchronization. The result of the multiple independent inverters varying real or reactive power without synchronization will produce an inadequate real power mismatch (∆P = ±20 p.u.) in the case of real power variation and the voltage at the inverter terminals remain within threshold. This method is effective in a single inverter connected to the grid case, but probability of failure increases in multiple inverter penetration. Also, large power variation might cause poor power quality. i.e. voltage flicker and grid instability. Impedance Measurement The impedance measurement method is derivative of the output variation method. The real output power is frequently varied while simultaneously 3 0 .88 ≤ V ≤ 1.1p.u. , 5 9.3 ≤ f ≤ 60.5Hz 6 1.2. Islanding Prevention calculating the network impedance by determining the rate of change of the inverter voltage with respect to the inverter current. An island is confirmed if significant increase in network impedance is observed above a predetermined impedance threshold. ∆V Z= (1.8) ∆I This method is similar to output power variation method but it is difficult to set a threshold for implementation. With local load varying over time, the threshold impedance could be rendered ineffective or cause inconvenient tripping of the inverter. For proper implementation and avoiding false detection, an accurate value of grid impedance must be arrived at which might not be known or available. Hence, this renders this method impractical [8]. Sliding Mode Frequency Shift (SMS) In this method, the grid voltage frequency is monitored and any increase in frequency from the nominal value will trigger the control of the DG inverter to increase the phase angle between the grid voltage and generated inverter current. In the absence of the grid, the increase in phase angle will result in a closer zero crossing and further increase in frequency and an uncontrolled slip occur until the frequency is no longer within nominal operating limits. The same concept is applied for any decrease in grid voltage frequency which will trigger the DG to further decrease the phase angle [9, 10]. In both cases, the voltage at the point of common coupling is monitored for frequency slip due to the unstable inherent nature of the detection method. This method is effective in detecting the islanding state and reduces the NDZ greatly except for a perfectly matched load which will maintain an unchanged frequency. To vary the inverter frequency, the starting phase angle of the inverter current is controlled via a sinusoidal function that is suggested for varying the starting phase angle of the current. π(f − f0 ) θ = θmax · sin (1.9) 2(fmax − f0 ) where θmax is the max phase shift corresponding to the max frequency change; change of fmax . In the presence of utility, it is assumed that inverter supplies constant real power output and zero reactive power. f0 is the utility frequency. 7 1.2. Islanding Prevention This method is efficient in reducing the NDZ close to zero. It provides a good compromise option between output power quality and accurate detection. This method would fail if the load phase angle and the starting phase angle are matched within the NDZ. Another possibility of unsuccessful detection is if the rate of change of the starting phase angle of the inverter is less than that of the load line with respect to frequency [8]. Active Frequency Drift (AFD) The method of active frequency drift is to push the inverter current frequency higher or lower by a positive feedback control design. Again, islanding is confirmed if inverter current frequency is pushed out of threshold limit [10]. The current waveform that implements a positive active frequency drift is shown in Figure 1.2. Since the inverter current is generated with a frequency slightly higher than that of the utility voltage, when the inverter current reaches zero it is held at zero for a chopping period Tef f until next zero crossing of utility voltage. Similarly, for the negative half cycle of the inverter current, when the current reaches zero, it is held for a period of Tef f until the utility voltage crosses zero again. When the inverter is connected to the Area EPS, reactive power is supplied to utility and can be expressed as Q = Vinv · Iinv · sin(ϕpf ) (1.10) where Vinv and Iinv are the rms inverter terminal voltage and current, respectively. ϕpf is the phase offset due to perturbing the frequency. Since the local EPS load is a parallel RLC load, the reactive power is expressed as ( ) 1 2 Qload = V −ω·C (1.11) ω·L and the reactive power mismatch can be expressed as ∆Q = Qinv − Qload (1.12) Therefore, it can be concluded that: • If (∆Q > 0) reactive power mismatch is positive, the frequency will increase gradually so the load can supply reactive power to match that supplied by the inverter and (∆Q = 0). • If (∆Q < 0) reactive power mismatch is negative, the frequency will decrease so the load produces less reactive power to match that supplied by the inverter and (∆Q = 0). 8 1.2. Islanding Prevention Figure 1.2: Voltage and current output of EUT implementing AFD This method of drifting the inverter frequency up/down is effective in detecting islanding and eliminating NDZ to almost zero. However, this method is rendered ineffective if the phase offset generated by perturbing the inverter frequency matches that of the load and within the NDZ region. Also, for widespread penetration, interaction amongst inverters implementing this method might render it ineffective and could cause detection failure. Except when all inverters adopt the same drifting direction. In Figure 1.2, when the island is formed, the voltage will change frequency matching the current and detection is achieved in principle. Main Monitoring Units with Allocated All-Pole Switching Devices (MSD) The principle of this method is to employ two separate mains monitoring with allocated switching devices (MSD) in cascade to perform a self test to ensure reliability of both islanding detection devices. The design of automatic disconnection devices allocated all-pole switches must be electromechanical with load break rating. i.e. relays or magnetic contactors. Any of the islanding methods can be employed in both MSDs. Over/under voltage and frequency can be employed. Impedance measurement method can be employed too. For grid impedance method, islanding is confirmed when a change in network impedance is observed [11]. Advantages and disadvantages of this method are similar to that of impedance measurement. In addition, the redundant design and automatic self test improves reliability of detection. Periodic approval of inverters requested by the distributed network operator (DNO!) is not required. The dispensable switches (SW2) in the redundant design adds additional cost to 9 1.2. Islanding Prevention the inverter. 1.2.3 Alternative Methods These alternative methods employ different techniques that are neither active nor passive methods. They are often employed at utility level and include: • Reactance insertion. • Carrier communication method. • Supervisory control and data acquisition. Reactance Insertion This method depends on the insertion of a low-value impedance to a distribution feeder with a short delay of time after disconnection [8]. If a capacitor bank is inserted a short delay after disconnection, it will supply additional reactive power to the load and unbalances reactive power between inverter and load. If this unbalance drives the island frequency out of threshold range, the newly created island is detected. Another variation of this method is the insertion of a low-value resistance to unbalance real power between inverter and load and drive the voltage out of threshold limits. This method is highly effective and reduces the NDZ to zero, if a capacitor bank is installed and coordinated with additional delay time. However, this method has a slow response time compared to active methods. Also, the cost for implementing this method is significantly higher as every disconnection switch need to be set up and equipped [6]. Power Line Carrier Communication (PLCC) The principle of this method is to use a low-power communication signal (beacon) sent by a transmitter at utility side through the distribution network. Islanding is confirmed if a receiver installed at the inverter can not receive the signal from the transmitter [12]. For a continuity test, a continuous communication signal is preferred since it is more reliable than a discrete or digital signal. With intermittent signals, loss of signal due to discontinuity or cessation of transmitter signal can not be distinguished without further encoding and decoding. Transmitter signal need to be of low frequency (i.e. < 500Hz) so it can pass 10 1.3. Standard Test of Unintentional Islanding inductance without difficulty. In addition, subharmonic signals are desired because loads are not are not able to reproduce it. This method is highly effective and efficient. It eliminates NDZ and is unaffected by the number of inverters connected to the utility. With existing automatic meter reading, a setup can be constructed to use AMR signal in conjunction with an inexpensive receiver. However, for now, this method is not as attractive due to the fact that this transmitter is uncommon and expensive. This method is most attractive for high density distributed generation penetration. Supervisory Control and Data Acquisition (SCADA) The principle of the SCADA method is to monitor the state of the entire distribution system (V, f, etc). When an inverter is installed, a voltage sensing device need to be installed for that part of the network [8, 13]. Through communication links with the control station, all sensing devices feed in their local information. If voltage is still sensed after utility disconnected from a particular part of the network, islanding is confirmed and measures are taken to cease generation in that part of the network to avoid personal injuries while servicing the isolated feeder. Out of phase re-closure can also be avoided. This method is highly effective and eliminates NDZ if proper implementation of instruments and control is applied. The cost of this method makes it unattractive as every inverter need to be fitted with a sensor. The cost of the sensor and communication to send information to central station makes this method impractical for the time being. With higher density of distributed generation and reduced cost of the instrument and communication in the future, might turn this method to be well suited for islanding detection amongst other applications. 1.3 Standard Test of Unintentional Islanding Present electrical power systems are designed to supply power unidirectionally. With more DG systems being connected to the EPS network, there are safety and power quality concerns that need to be addressed. A uniform standard of interconnection was established in the form of IEEE Std 1547 - IEEE Standard for Interconnecting Distributed Resources with Electric Power Systems. Of particular interest to this work is the test procedure for unintentional islanding test outlined in IEEE Std 1547.1-(5.7.1) [14]. 11 1.3. Standard Test of Unintentional Islanding Figure 1.3: Initial islanding test circuit with matched load Figure 1.4: Islanding test circuit according to the German proposal [15] Islanding condition of matched real power between DG inverter and local EPS load to obtain as close as possible zero current from the utility is sometimes sufficient for the creation of a stable island [15]. For newer inverters with unity power factor and no islanding detection control the probability of islanding is greatly increased. Figure 1.3 shows an earlier islanding test circuit used to island the test DG inverter [15]. The procedure was to vary the resistive load to obtain as close as possible zero current from the grid side. As inverter technology advances, more inverters with low frequency ringcore transformers are manufactured. These inverters inherently draw a certain amount of reactive power from the grid. Therefore, an earlier German proposal [15] suggested the addition of a resonant LC elements to simulate the mains with limits of ±100 VARs in addition a resistive impedance in parallel with switch SW as in Figure 1.4. The matching of real and reactive power with the load increases the probability of islanding especially if the load resonant frequency is tuned to the EPS frequency. The German proposed test added discrete island stabilizing elements, L and C, which made it closer to actual islanding conditions. (EUT). Another islanding test circuit was proposed in Japan that is similar to the German proposal except with the addition of an idling ac motor of around 500W and without the mains simulated oscillator circuit and Z = ∞ [16]. The standard for interconnecting DR with electric power systems was 12 1.4. Advantages and Shortcomings of the Standard Test Figure 1.5: IEEE Std 1547 unintentional anti-islanding test circuit developed due to a growing concern for the probability of unintentional islanding and to have a uniform benchmark for interconnecting fuel cells, photovoltaics, distributed generation, and energy storage units with electrical power systems [1]. The standard requires all equipment meant for connecting with the grid EPS to pass an unintentional anti-islanding test. The EUT must detect the simulated island of the circuit in Figure 1.5 within a specified time limit (2 sec) to be certified as a non-islanding inverter appropriate for interconnection with the grid systems. 1.4 Advantages and Shortcomings of the Standard Test For an unintentional islanding test, the reactive-power portion originating from the active frequency drift for example is not compensated for by the test LC load [13]. That means reactive power mismatch typically is not zero and hence implies that the conditions of the test do not simulate a worst-case scenario. In addition, the case where other detection methods dynamically perturb or indirectly change the inverter output reactive power, the static LC load is not capable of tracking changes to reactive power mismatch. The same applies for methods based on voltage or real power variation. The static RLC load in the unintentional islanding test poses the question of how effective will the static RLC load be in an islanding test with more advanced DG systems being developed? The island RLC load is clearly effective in matching the bulk power generated by the DG unit under test but there is a clear need for a dynamic element to guarantee matching P and Q of the load to that of the equipment being tested. Another shortcoming of the discrete simulated island load is the fact that cost, size, test environment and setup time are factors that need to be considered when testing higher power inverters. Due to the unity quality 13 1.5. Motivation and Dissertation Outline factor required in the test, inductance and capacitance size increases dramatically. As an example, a quality factor of 0.5 would lead to a significant value for inductance and capacitance [17]. 1.5 Motivation and Dissertation Outline Initially, approaching the islanding issue from the probability of its occurrence has raised some serious questions of how valuable the research would be in this topic. As reviews of the problem take shape considering future implication of higher penetration and distributed generation, the islanding problem becomes a key topic in the interconnection of DG inverters with the EPS network. Researching the methods for islanding detection has motivated this research initially to investigate the possibility of improving some already established detection methods by reducing their non-detection zone. Active frequency drift was one method that was investigated as well as negative sequence current injection. As the research and researcher matures in the topic of islanding, focus was shifted to how these methods are evaluated. The numerous islanding detection methods approach islanding detection in different and some in radical ways. From passive measurement and analysis, to actively perturbing system parameters, to using carrier signals to exclusively determine the state of the local EPS network. The accurate representation of the lab simulated island is key to effectively determining the actual NDZ of the islanding detection method under test. The motivation for this research stemmed the problem of how to accurately represent the simulated island in order to reproduce a worst case islanding scenario in a lab environment that meets the criteria of IEEE 1547 and be accurate, cost effective, reproducible, scalable, and portable. The thesis first reviews the islanding problem and methods of detecting islanding conditions. In chapter 2, the unintentional islanding test of IEEE 1547 is presented and simulated. The results of the test will be discussed and a test bench is established for accurate comparison between the proposed islanding test and the standard one. Next, the proposed unintentional islanding test is introduced in all it’s derivatives in chapter 3. An initial and a more robust designs of the ISE are presented in chapter 4 for single and three phase circuits. Chapter 5, will include results of the proposed unintentional islanding test and a comparison to the standard test. Experimental results of the development of the ISE is presented in chapter 6 and finally conclu14 1.5. Motivation and Dissertation Outline sion of the work done and results are summarized in chapter 7 in addition to future work. 15 Chapter 2 The Standard Unintentional Islanding Test Islanding condition of matched real power between DG inverter and Local EPS load to obtain as close as possible to zero current from the utility is sometimes sufficient for the creation of a stable island [15]. For newer inverters with unity power factor and no islanding detection control the probability of islanding is greatly increased. The standard for interconnecting distributed resources with the electric power systems was developed as a result of the growing concern regarding the probability of unintentional islanding and to have a uniform benchmark for interconnecting fuel cells, photovoltaics, distributed generation, and energy storage units with the electrical power systems [1]. The standard requires all equipment meant for connecting with the grid EPS to pass an unintentional islanding test. The equipment under test (EUT) must detect the simulated island in the circuit of Figure 2.1 within a specified time limit to be certified as a non-islanding equipment appropriate for interconnection with the grid EPS. According to IEEE Std-1547.1, an EUT passes the unintentional islanding test if detection occurs within a 2 sec window beginning from the time of the simulated island creation [1]. The unintentional islanding test of IEEE’s Std. 1547 is used here to measure the effectiveness of the proposed method. Since the simulated island Figure 2.1: Unintentional islanding test per-phase circuit 16 Chapter 2. The Standard Unintentional Islanding Test is represented by a parallel RLC load, the power delivered to the simulated island is governed by Pload = V2 R[ Qload = V 2 1 − ωC ωL ] (2.1) (2.2) where Pload and Qload are the real and reactive power delivered to the load, respectively. In the presence of the utility and assuming the EUT operate at unity power factor, real power will be governed by the voltage at PCC as per Equation 2.1. Since the EUT is programmed to supply the full load real power at utility voltage, voltage imbalance after utility disconnect will be minimized and would remain within nominal operating limits. On the other hand, in the absence of the utility, the voltage at PCC is unregulated and any mismatch in real power will either drive the voltage up or down to achieve real power equilibrium between the EUT and the simulated island load (R). In the same circumstance, any reactive power mismatch will drift the island frequency to the simulated island (LC) resonant frequency of Equation 2.2 to achieve reactive power equilibrium. In order to establish the simulated island, a matched resonant LC load tuned at utility frequency will ensure maintaining the frequency within nominal operating limits after griddisconnection. Also, to maintain the voltage within threshold limits, real power generated by the EUT should be as closely matched to the simulated island load (R) to maintain nominal voltage at PCC. These two conditions ensure sustaining the island and maintaining normal voltage and frequency operating conditions. For the circuit of the unintentional islanding test of Figure 2.1, the RLC elements could be calculated based on the following equations: R ωQf Qf C= ωR √ √ C QL QC Qf = R = L P L= (2.3) (2.4) (2.5) where Qf is the quality factor for the parallel RLC load and QL and QC are the reactive powers for L and C, respectively. According to IEEE 1547 and the European standard for interconnection IEC 62116, Area EPS is typically operated at a power factor higher than 17 2.1. Simulation Environment 0.75 under normal conditions. In the unintentional islanding test, a unity load quality factor would ensure testing the EUT under near real-world conditions. In the European standard a load quality factor of 0.65 is specified for the unintentional islanding test which would imply an uncorrected line power factor of 0.84. The lower the load quality factor, the easier it is to design perturbation based active methods that do not influence power quality negatively. Requirements of the IEEE Std. 1547 unintentional islanding test stipulate a Qf of 1, an equivalent to an uncorrected line power factor of 0.707, which is a closer representation of real-world line power factor for a possible island as opposed to the wider range of IEEE Std. 929-2000 load quality factor of ≤2.5 that represents an uncorrected line power factor of 0.37 or higher [2]. This islanding test setup and parameters conditions ensures minimal power mismatch supplied to or delivered from the grid EPS. In this chapter, The simulation tool used to build the models is described and the standard unintentional islanding test circuit is detailed. Then the test circuit parameters are calculated and a verification of the proper performance of the circuit is conducted to ensure the creation of a sustainable island. After which unintentional islanding tests are carried out for both the single and three-phase inverters referred to them as EUTs. Finally, analysis of unintentional islanding tests are discussed and possible improvements are suggested. 2.1 Simulation Environment The simulation software that was most appropriate for this research is PSIM [18]. It is a power electronics simulation software with the capability of fast simulation and of switch based circuits. Also it offers ease of use within powerful simulation environment. The complex design of the islanding test becomes apparent with an accurate design of the non-islanding inverter, single or three phase. PSIM offers an intuitive and straight forward graphic user interface for the design. The large design of the proposed island stabilizing element, the inverter model and control of both inverters, became an evident issue with other simulation environments. PSIM provided a subcircuit container which made it possible to compartmentalize distinct parts of the circuit for ease of tracing, debugging and the power of scalability and reproducibility in other simulations. PSIM is efficient for simulations of circuits with convergence problems and long simulation time. Its simulation engine is faster than other simula18 2.2. Test Circuit SW3 I_util I_EUT VPCC A SW2 A SW1 Simulated Island I_Load EUT o Area EPS IC IR IL Figure 2.2: Per-phase unintentional islanding test circuit. tion software and it allows for repetitive simulation runs reducing the design cycle. PSIM’s unique features in the handling of power converter circuits, control circuits, and system integrated simulation with MATLAB through SimCoupler made it the ideal simulation software to be used to simulate the majority of the simulation for this research. 2.2 Test Circuit The outline of the islanding test circuit is illustrated in Figure 2.2. If the resonant LC load in the figure is matched at a resonant frequency equal to the power system frequency, that will ensure minimal fluctuation in operating frequency at the the moment of island creation (grid-disconnection). Also, when real power is balanced between load and EUT, it would maintain the voltage at PCC within nominal operating limits. 19 2.3. Simulated Island Parameters Table 2.1: Unintentional islanding test parameters Parameter Single-phase Pinv Qinv Vn Iinv fo fs 4 R L C PF Qf 1 0 120 8.33 60 10 14.4 38.179 184.207 1 1 kW kVAR Vrms Arms Hz kHz Ω mH µF According to IEEE 929-2000, a simulated island load with a load to generation ratio between 50% and 150% and resonant frequency of 60Hz/50Hz with quality factor Qf = 2.5 or less will satisfy all distribution line power factors and will provide a lab simulated environment to test single inverters unintentional islanding detection [2]. In the more recent IEEE standard of 1547, the quality factor is set to Qf = 1 ± 0.05 as the simulated island quality factor. It is found to be more realistic to covers most distribution line power factors and provide closer to worst case conditions [1]. 2.3 Simulated Island Parameters In Figure 2.2 the layout of the standard unintentional islanding test required in IEEE Std 1547 was shown. To calculate the size of the simulated island load, the EUT power rating and power factor, Area EPS nominal voltage and frequency need to be specified. Simulated island load can be calculated using Equation 2.3, Equation 2.4 and Equation 2.5. Table 2.1 shows the unintentional islanding test specifications and the simulated island parameters {R,L,C} for both single- and three-phase PWM EUT. 20 2.4. Test Procedure 2.4 Test Procedure The standard test applies for both single and three-phase systems. The simulated island load (RLC) is calculated per-phase. Therefore, for the three-phase EUT test, there will be three identical sets of simulated island loads assuming a balanced testing conditions. The initial test starts at the balanced condition between load and EUT with load quality factor set to one. Once the results of this stage are obtained, the reactive load is adjusted between 1 ± 0.05 in 1% increments from the initial balanced load condition. It is possible ,if convenient, to adjust the EUT output reactive power to produce 1% increments up to ±5 from the initial balanced condition. If islanding detection time remain increasing, further test iterations of 1%increments of reactive power are performed until the detection time begin to decrease. The next stage of the test reviews the results and two more test iterations are performed for the three longest trip times with the same 1% increments. Next, the test is repeated for 66%5 and 33% EUT output power settings if permissible. 2.5 Simulation Results of the Standard Unintentional Test The standard islanding test was carried out for five EUT cases. In the first case the EUT tested did not implement any islanding detection scheme in the control. The purpose of this experiment is to verify the parameters chosen for the simulated island (RLC load) and the proper performance of the islanding EUT in maintaining the island during the test. In the second and third cases the standard islanding test was carried out using an EUT implementing an active frequency drift(AFD) and reactive power variation (RPV) anti-islanding control schemes, respectively. The fourth and fifth cases tested a three-phase EUT with no islanding detection and with RPV anti-islanding control scheme, respectively. 2.5.1 Equipment Under Test (EUT) The Single-Phase EUT: is a single-phase 1kW/120V inverter switching at 20kHz connected to a 300V DC bus. The control implemented is a 5 50% to 95% output power settings are allowed 21 2.5. Simulation Results of the Standard Unintentional Test simple PI-control and phase-lock loop (PLL) to maintain unity power factor operation. The Three-Phase EUT: is a three-phase 3kW/208V inverter switching at 10kHz connected to a 300V DC bus. The control implemented is a synchronous frame space vector modulation control and the unit is operating at unity power factor. Note: The EUTs power rating used was not due to limitations in the design but to illustrate a per unit rating. The single phase EUT could be scaled (10kW, 100kW, 200kW, etc.) and the same respectively applies for the three-phase EUT. 2.5.2 Test Start/Stop All simulation was carried out for the circuit in Figure 2.3a for single-phase and in the case of three-phase EUT the same circuit is used per-phase. During the test, the Area EPS disconnection occur at t=0.5 sec and the simulation ends at 3.5 seconds. The initial 0.06 seconds are omitted as it is a transient period of the simulation that is not part of the test. In order to determine a PASS or FAIL result, the EUT voltage and frequency Over/Under trip signal is added to the figures respectively to show proper islanding detection if the trip signals occur within the 2 seconds test window from the time the Area EPS disconnects. 2.5.3 Case 1: The Single-Phase Islanding EUT An islanding EUT is connected to the test circuit to verify creating and sustaining the island within nominal operating voltage and frequency limits. In Figure 2.3a, the magnitude of the PCC voltage before and after grid-disconnection is shown to remain within the upper and lower voltage limits for the period of the test. The minimal change in voltage amplitude indicates that real power matching condition is achieved. For the frequency of the EUT, Figure 2.3b shows the frequency also with in upper and lower limits. The frequency remaining virtually constant before and after griddisconnection is an indication that the reactive power matching condition is also achieved. These two parameters are enough to successfully establish and sustain the simulated island. Figure 2.4 illustrates the EUT output current and PCC voltage before and after grid-disconnection. The control for this EUT samples the PCC 22 2.5. Simulation Results of the Standard Unintentional Test (a) Voltage magnitude (pu). (b) Frequency (Hz). Figure 2.3: Voltage and frequency 23 2.5. Simulation Results of the Standard Unintentional Test Figure 2.4: EUT, EPS, and load currents (pu). voltage and calculate the reference current according to the command power then synchronize the current to PCC voltage via PLL for unity power factor operation. During this test, real and reactive power are monitored and the expected mismatch in this case is zero. Power-matching of the EUT and the simulated island (RLC) in the presence or absence of the Area EPS is expressed as Pload = PEU T + ∆P (2.6) Qload = QEU T + ∆Q (2.7) The EUT, Area EPS, and load real and reactive powers are shown in Figure 2.5a and Figure 2.5b. The simulated Area EPS real and reactive power flow immediately before the moment of grid-disconnection is almost zero (P̄EP S = 0.5 × 10−3 pu , Q̄EP S = 0.06 × 10−3 pu). While the EUT and load are matched at 1pu of real power and since the load LC are matched, the apparent load is resistive and the EUT operating at unity power factor therefore the reactive power produced by the EUT is also negligible (Q̄EP S = 0.39 × 10−3 pu). The compared results confirm the matching power condition and confirm the ability of the EUT to sustain the simulated island throughout the test period. Continuing the test for the purpose of illustrating the test circuit proper performance, the reactive EUT power was incremented by 1% to ±5% from initial balance condition (QEU T = 0). Trip times have been recorded as 24 2.5. Simulation Results of the Standard Unintentional Test (a) Real power (pu). (b) Reactive power (pu). Figure 2.5: Real and reactive power . 25 2.5. Simulation Results of the Standard Unintentional Test Figure 2.6: Frequency response of EUT during the standard islanding test. 26 2.5. Simulation Results of the Standard Unintentional Test shown in Figure 2.6. In this case, the EUT fails to detect islanding between +2% and -1% increments of reactive power. For negative reactive power increments, the trip times are noticed to reduce while the positive increments increase between 3% and 4% and then reduce at 5%. This EUT failed the standard test since no islanding detection was implemented and it is noted that the NDZ for this device is roughly within +3% and -2% of reactive power mismatch, the EUT over and under frequency controls would detect islanding. 2.5.4 Case 2: The Single-Phase EUT with AFD Figure 2.7: EUT PI-controller with AFD control scheme. In this experiment the EUT control implements an active frequency drift scheme for islanding detection as shown in Figure 2.7. The control implemented is similar to that reviewed in chapter one. The control of frequency drift is achieved through the chopping fraction expressed as CF = tz 0.5TEP S (2.8) where TEP S is the Area EPS frequency and tz is the time where the EUT current remain zero until the Area EPS voltage reaches zero again (See Figure 2.8). The EUT frequency could be calculated as fEU T = 1 TEP S − tz (2.9) 27 2.5. Simulation Results of the Standard Unintentional Test and the drift in frequency is the difference between the Area EPS and the EUT frequencies. In the first part of this case, the chopping factor was set to a constant +5% representing a constant current command frequency of 63Hz. Figure 2.8: EUT current with AFD implementation Figure 2.9 shows the frequency response of the EUT to the 1% increments in load mismatch in reactive power starting from the balanced condition of matched load to ±5%. It is clear from this graph that the EUT manages to detect islanding for these conditions. Figure 2.10 presents the frequency out of nominal limit trip times. The three largest trip times occur at 0.97, 0.96, 0.95 of load reactive power, hence requiring further 1% increments until trip times begin to reduce. Further 1% increment of load mismatch in reactive power reveal increasing trip times. At 0.94QL ttrip = 0.02479 and at 0.93QL ttrip = 0.03314. However, at 0.92QL the EUT fails to detect the island and fails the test. The next part of this case involves adjusting the AFD scheme to be able to detect islanding. The adjustment was made to the chopping fraction so that is varies every Area EPS cycle from +5% to -5%. This simple adjustment ensures the ability to detect islanding within the allowed window. results of this part of the case is summarized in Figure 2.11. The 1% increments in load mismatch in reactive power had to be continued to 0.88QL to observe a decrease in trip times. Hence, this EUT passes the first part of the standard islanding test. The test requires, if possible, that the EUT output power be adjusted to 28 2.5. Simulation Results of the Standard Unintentional Test Figure 2.9: Frequency response for EUT under standard isalnding test Figure 2.10: EUT trip time for AFD chopping fraction of 0.05 29 2.5. Simulation Results of the Standard Unintentional Test Figure 2.11: EUT with modified AFD trip times (CF=±5%). 66% and 33% and repeat the procedure. For the purpose of this research, this portion of the test will not be carried out since it would not add more value for the comparative analysis with the to be proposed test adjustment. 2.5.5 Case 3: The Single-Phase EUT with RPV Reactive power variation is implemented in the EUT control. The current reference is manipulated through a digital PLL [19] to generate a periodical (0.5 s) change in EUT reactive power in an attempt to drive the frequency out of nominal limits. This method was also described in chapter one. An implementation of this method is illustrated in Figure 2.12 which shows the current control loop. In the first stage, Figure 2.12a, the PCC voltage is used to generate an in-phase and 90 degree shifted reference signals labeled PLL sin and PLL cos, respectively. In order to generate the required in-phase and 90o phase shifted sinusoidal, β is required and calculated as β= −1 + 0.5aTd 1 + 0.5aTd (2.10) where a is the inverse of the all pass filter time constant and Td is the discrete time step. The next step in this method is to manipulate the output signals of the digital PLL and scale them according to the real and reactive power commands. For unity power factor EUT, zero reactive current is commanded, yet for RPV the reactive power current reference is slightly adjusted to generate a periodical reactive power variation with zero average over 0.5 seconds. Figure 2.12b shows the current reference loop for the EUT. Qpct 30 2.5. Simulation Results of the Standard Unintentional Test (a) (b) Figure 2.12: RPV current control loop is the percent reactive power for islanding detection. In this case, reactive power variation is chosen as ±5%. The period of the reactive power variation of 0.5 seconds was chosen so that within the required 2 seconds window at least two detection cycles are possible and reducing the NDZ as a result. The results of the standard islanding test for an EUT with reactive power variation are shown in Figure 2.13 for a matched load. Islanding detection for this iteration of the test occur at 0.05041 seconds. The test’s test results are summarized and shown in Figure 2.14. It is can be observed that trip times are decreasing as mismatch in reactive power increase up to 2% mismatch where the longest trip time is recorded. This test case provide a successful islanding detection by the EUT under standard islanding test procedure. 2.5.6 Case 4: An Islanding Three-Phase EUT The 3kW-120V inverter was connected to the simulated island RLC load and run at 100% rating for 2.5s. The grid-disconnection occurred at 0.25s. Figure 2.15a shows the load active power, and the active power supplied by the inverter and the grid before and after grid-disconnection. It is clear to depict that the inverter active power (Pinv ) is almost matching that of the load (Pld ) before and after grid-disconnection. Figure 2.15b illustrates the inverter reactive power (Qinv ), grid reactive 31 2.5. Simulation Results of the Standard Unintentional Test Figure 2.13: EUT with RPV islanding test Figure 2.14: EUT trip times 32 2.5. Simulation Results of the Standard Unintentional Test (a) Active power (b) Reactive power Figure 2.15: Real and reactive power of the inverter, load and grid power mismatch (∆Q), and one of the load passive element’s (L or C) reactive power. The load quality factor of 1 required by the standard islanding test could be verified from Figure 2.15b. The active and reactive power mismatch between the island load and the inverter supplied by the grid is shown to be almost zero supporting a worst case islanding scenario and the unity power factor operation of the controlled inverter. The voltage and frequency at PCC before and after grid-disconnection are shown in Figure 2.16. From this figure, it can be noticed that the voltage and frequency are within nominal operating limits after grid-disconnection for over 2s confirming a complete islanding creation. (a) Inverter voltage amplitude (b) Inverter voltage frequency Figure 2.16: PCC voltage magnitude and frequency 33 2.5. Simulation Results of the Standard Unintentional Test Figure 2.17: Inverter voltage and current before and after grid-disconnection 2.5.7 Case 5: A Three-Phase EUT with Negative Sequence Injection This case is based on generating an unbalanced reference current commands for the space vector modulation. The synchronous frame Vd∗ and Vq∗ of the reference current control are transformed into rotating frame and injected with a set of three 1% perturbation shifted by 120o at each phase zero crossing. Then the modified rotating reference current commands transformed back to synchronous frame for space vector modulation processing to generate the modified gate signal commands for the three-phase inverter. The objective of this test is to determine the effect of unbalance in inverter currents on the inverter terminal voltages and consequently PCC voltages. In the case of grid connection, the voltage of the PCC is governed by the utility voltage and no significant effect is noticed. In the absence of the grid and continuous stable islanding operation by the inverter, the voltage of the PCC will be affected by the current unbalance due to negative sequence current injection. By monitoring the negative sequence component of PCC voltage, islanding detection could be observed and cease to energize command is issued in the presence of negative sequence PCC voltage above a predetermined threshold level. In Figure 2.18, the instantaneous negative sequence component in the PCC voltage under islanding operating conditions without the propose current space vector reference command perturbation. Also, it shows the same component after perturbing the reference current SVM voltage commands by 1% and 2%. From this figure, it is worthy to point that the instantaneous 34 2.5. Simulation Results of the Standard Unintentional Test negative sequence component at PCC is almost unity with grid connection even with 1% or 2% of negative sequence current injection. After griddisconnection, 2% and 6% change in negative sequence component appears at 0% and 2% injection. This change in negative sequence component is enough to detect the islanding condition. While it is widely assumed that the grid is naturally balanced, there is an inherent minimal level of unbalance that would enable the negative sequence threshold level to be determined based on this reference negative sequence voltage. In recent studies, the negative sequence voltage of a 100MVA, 600V bus system ranged from 0.14 to 0.63V while the negative sequence current ranged from 1.0 to 11.7A [20]. Figure 2.18: Negative sequence voltage component at PCC 35 Chapter 3 Proposed Island Stabilizing Element (ISE) 3.1 Overview of Advantages and Disadvantages of Standard Islanding Test In the previous chapter, the standard islanding test was demonstrated. It was concluded that there are limitations to the current standard test which are summarized in the following: 1. The RLC load is a static load, hence any changes during the test are not compensated. 2. The RLC load tend to increase dramatically in size with higher EUT power rating. 3. The cost for the RLC and setup cost will increase with the power rating of the EUT. 4. The RLC load does not monitor the PCC for any mismatch in power on the EPS side. A more accurate unintentional islanding test is desired. A test that addresses the previous points and recreates in a lab controlled conditions worst case islanding scenario. For reference, the circuit of the unintentional islanding test is shown in Figure 3.1. In the next sections, three islanding test topologies will be introduced and an initial design for the island stabilizing element proposed will be discussed. 3.2 Modified Unintentional Islanding Test The modification proposed for the islanding test is intended to address the issues discussed in the previous section. In order to eliminate the size issue with larger power EUTs, a power electronic solution is proposed to replace 36 3.3. Island Load P & Q Figure 3.1: Standard unintentional islanding test the RLC load of Figure 3.1 fully or partially. A current controlled current source is used as an island stabilizing element, their after labeled ISE. The main function of the ISE is to monitor the voltage at the point of common coupling and current of the EUT and match the active and reactive power so that the mismatch in P and Q is reduced to zero. In this method, a worst case scenario for islanding is achieved while maintaining a relatively contained size for the load while reducing cost and setup time. In addition, the ISE makes it possible to dynamically match the EUT real and reactive power during the test so that any mismatch produced during the test is matched by the load. In the following sections, an initial investigation into different proposed unintentional islanding tests topologies is carried out. The first topology is replacing the resonant elements with the ISE (representing Qload only) and analyzing the performance of the island. The second topology is a more encompassing approach by letting the ISE represent the complete island load (producing Pload and Qload ). The last topology option places the ISE in parallel with the bulk island load (RLC). In this setup, the ISE will function as a P and Q mismatch compensator during the test. These different topologies have advantages and disadvantages that will be discussed in each section accordingly. 3.3 Island Load P & Q Referring to the standard test circuit of Figure 2.1, Equation 2.6 and Equation 2.7, repeated below for convenience, hold true for any of the proposed unintentional islanding test topologies. Pload = PEU T + ∆P (3.1) Qload = QEU T + ∆Q (3.2) 37 3.4. ISE Representing Qload while within the island load, Pload and Qload are dependent on the function of the ISE and its location. A general form of the load active and reactive power can be arrived at when considering all possible elements. Pload = PR ± PISE (3.3) Qload = QL − QC ± QISE (3.4) The ISE real and reactive power sign will depend on the amount of mismatch in power between the load and the EUT in the presence of the area EPS. After area EPS disconnects, the voltage of the island will be in sync with the EUT current and the amount of reactive power mismatch after area EPS disconnection will force a frequency shift. 3.4 ISE Representing Qload Figure 3.2: Modified island load with R and ISE The ISE is used to replace the function of the LC components of the simulated test island. Figure 3.2 shows the proposed circuit. The ISE will be responsible for representing the LC elements internally and producing current only in specific reactive power mismatch conditions. Under these conditions, the ISE will produce a current with appropriate magnitude and phase to eliminate any mismatch in reactive power or generate an intentional mismatch in reactive power according to the demanded quality factor through out the test. It is important to note here that there are two kinds of reactive power mismatch conditions. The first is produced by the EUT during the test, i.e. due to islanding detection method implemented or none unity power factor operation. The second mismatch in reactive power is an intentional one that is part of the unintentional islanding test. In the later case, the ISE is required to produce reactive power mismatch in increments of 1% of the load quality factor6 . 6 Qf range: 1 ± 0.05 38 3.5. ISE Representing Pload & Qload Therefore, in the case of matched LC Equation 3.4 will be the same as Equation 3.2 and that will maintain zero reactive power sourced from the EPS side (∆Q = 0). On the other hand, since the ISE represents only the LC elements, PISE = 0 and PR has to be manually matched to PEU T to insure that there is no mismatch in active power. The ISE unit current rating will depend only on the amount of mismatch in reactive power of the LC elements in addition to any reactive power from the EUT. For 3-phase 100KVA 480V EUT, the inductive reactance is compensated by the reactive reactance simulated within the ISE so that only the amount of mismatch between them need to be generated. Therefore, for a unity power factor EUT with no change in voltage or frequency, the ISE current is ideally zero. The disadvantages in this setup is that the real power is wasted as heat in the resistor. With higher power EUTs the resistor bank becomes less economical and could be replaced by the ISE. 3.5 ISE Representing Pload & Qload Figure 3.3: Modified island load with ISE only As in the previous section, the ISE will replace the function of the RLC in the unintentional islanding Test. Figure 3.3 shows the layout of the circuit. As described before, the ISE will simulate the function of the LC elements in addition to representing R. The ISE will function as a real power dump load by matching the EUT sourced active power and either storing it into a battery management system or converting it in multistage back to AC and injecting it into the EPS network. In this case, rather than exhausting power as heat, a power management system ensures the recycling of the power generated by the EUT and minimizing the wasted energy almost to zero. This setup will be optimum for low to mid power rated EUTs but requires design considerations for higher power units with regards to storage capacity and power recycling back to 39 3.6. ISE Representing Mismatch in P and Q the EPS network. 3.6 ISE Representing Mismatch in P and Q Figure 3.4: Addition of ISE to island load In this last configuration, as illustrated in Figure 3.4, the ISE will monitor the load and EUT for any mismatch in power and will source or drain power accordingly. For the situations where the available RLC bank is not sufficient to match the EUT rated power, the ISE would supply the difference in power as in PISE = PEU T − PR (3.5) QISE = QEU T − QLC (3.6) which will result in fixing the size of the RLC load. In this setup, the ISE rated current will be dependent on the size of available RLC load used in the test. The advantages of this test setup rather than the previous ones is the fact that both P and Q are monitored by the ISE and only the difference in power would be supplied by the ISE. On the other hand, the design of the ISE needs prior knowledge of the size of RLC load used in the test. Also, the size of the test equipment is relatively larger than the case of representing only the reactive loads. 3.7 Initial Design of the ISE Since the island load in the islanding test is applied per-phase, the ISE unit design is based on a current controlled current source and the single phase circuit diagram is shown in Figure 3.5. The ISE circuit topology in 40 3.7. Initial Design of the ISE Proposed Test Island s1 Vdc s2 rL L R EUT s3 Area EPS C Figure 3.5: Proposed test island stabilizing element this figure shows the case of replacing the LC elements and chosen in this section to illustrate the design and control of the unit. The process of modelling of the island stabilizing element to replace the island load in the standard test was in two stages. Initially, a small-signal based model of the ISE was developed. Later, a synchronous reference frame model of the ISE was developed. The reason for developing the two models is the different control strategy that could be applied and improvement in control. The circuit of Figure 3.6 shows a per-phase representation of the ISE. Here, the single phase inverter could be represented as a sinusoidal voltage source connected to the utility via an L filter considering fundamental frequency only and neglecting the higher switching frequencies [21]. Figure 3.6: Per-phase circuit diagram Where ua , ia and EPS (va ) are the PWM inverter voltage, inductor current and utility voltage, respectively. 41 3.7. Initial Design of the ISE 3.7.1 Small-Signal Model of the Single-Phase ISE To derive the transfer function of the ISE, a small-signal circuit analysis is carried out to obtain the input-to-output and control-to-output transfer functions. The small-signal analysis was carried out for one switching period, Ts , and the complete (DC & AC) mathematical representation of Figure 3.6 is expressed as d ⟨ia ⟩ dt d d′ : va = −ua + rL · ⟨ia ⟩ + L · ⟨ia ⟩ dt d : va = ua + rL · ⟨ia ⟩ + L · (3.7) (3.8) It is noted here that ⟨ia ⟩ is the average of the inverter current over a switching period and while in Figure 3.6 the inverter is not explicitly shown to reverse polarity during d′ Ts , it is implied to reverse polarity when switches change configuration from state dTs to d′ Ts . In Equation 3.7 and Equation 3.8 ua , va , ⟨ia ⟩, d and d′ are defined as follows u a = Ua + u ba va = Va + vba ⟨ia ⟩ = Ia + bia (3.9) d = D + db d′ = 1 − d Where the term accented with a (b) is a small signal variable and capital terms denote a quiescent variable. Equation 3.7 and Equation 3.8 are added to obtain the average inverter voltage over one switching period. d ⟨ia ⟩ dt d = (2d − 1)ua + rL ia + L ⟨ia ⟩ dt va = (d − d′ )ua + rL ia + L (3.10) Replacing the variables in Equation 3.10 with their quiescent and small signal terms of Equation 3.9 yields a full expression of the single-phase PWM inverter. d Va + vba = (2D + 2db − 1)(Ua + u ba ) − rL (Ia + bia ) − L (Ia + bia ) dt (3.11) 42 3.7. Initial Design of the ISE Considering only the small signal terms of Equation 3.11 to obtain the input-to-output and control-to-output transfer functions, the equation will be expressed as b a − rLbia − L d bia vba = (2D + 2db − 1)b ua + 2dU dt (3.12) Taking the Laplace transformation of Equation 3.12 yields ba (s) + 2D(s) b U ba (s) + 2Ua D(s) b Vba (s) = (2D − 1)U − (rL + sL)Iba (s) (3.13) The input-to-output transfer function of the single-phase PWM inverter is arrived at by setting the small signal utility voltage, Vba (s), and perturb ba (s)| bation in duty cycle, D(s), to zero and solving for G(s) = |Iba (s)/U yields 2D − 1 (3.14) rL + sL The control-to-output transfer function could be obtained by following the same procedure of setting the small signal utility voltage, Vba (s), and ba (s), to zero and solving for G(s) = |Iba (s)/D b a (s)| yields inverter voltage, U G(s) = Gdb(s) = − 3.7.2 2Ua rL + sL (3.15) PI-Controller Design A PI controller , Gc (s), is implemented to improve the control system performance and provide compensation for the single pole Gdb(s) transfer function. The traditional PI compensator transfer function is expressed as Gc (s) = Kp · sTI + 1 sTI (3.16) where KI = Kp /TI . Figure 3.7: Closed-loop PI-Control system In Figure 3.7 the compensated closed-loop control system is presented. In order to translate the output of the PI-controller to the input of the 43 3.7. Initial Design of the ISE control-output transfer function Gdb, a comparator is used to generate the duty cycle after a comparison of the output of Gc with a saw-tooth function. Figure 3.8 shows the two inputs of the pulse-width modulator, Vsaw and Vc . From the graph, the linear relationship between the output of the PI controller to the duty cycle is evident and the transfer function of the pulsewidth modulator is found to be 1/VM . Figure 3.8: Generation of the pulse width modulated signal 3.7.3 PI-Controller Analysis Figure 3.9: Generation of the pulse width modulated signal Using PSIM software the closed loop control-to-output loop was implemented as in Figure 3.9. Values for the PI controller were arrived at using iterative trials. Optimum values for Kp and Ti were found to be 7.785 and 0.0001, respectively. a saturation block was used to limit the PI output between ±2.48 and VM was set to 2.5 at switching frequency of 10kHz. Also, a unit step response was obtained for the control-to-output loop as in Figure 3.10. It can be seen that the percent overshoot is within 10% and the response of the system is critically damped. A bode plot of the open-loop control-to-output TF is shown in Figure 3.11 where it confirms the stability of the system with the phase magnitude is well below the 0 degrees. 44 3.7. Initial Design of the ISE Figure 3.10: Generation of the pulse width modulated signal Bode plot of Control to Output TF −1 10 0 10 1 10 Frequency (Hz) 2 10 3 10 4 10 5 10 Figure 3.11: Bode plot of control to output TF 45 3.7. Initial Design of the ISE 3.7.4 Average and Ripple Current To obtain the average and ripple currents in the output stage of the ISE inductor L in Figure 3.6, a list of assumptions are specified as below: 1. During one switching period, the area EPS voltage could be assumed constant and is replaced with a DC voltage source. 2. The ripple is calculated over the window dTs . 3. For values of simulated impedance Zpq ≫ rL , voltage drop across rL could be ignored. 4. For simulated reactive loads, the maximum inductor ripple occur when the area EPS voltage va is at zero. The ISE inductor voltage is defined as vL = L dia dt (3.17) also, in terms of the Equation 3.7, the inductor voltage is found to be equal to vL = ua − va + vr (3.18) From Equation 3.17, Equation 3.18 and ignoring vr , the peak-peak current ripple can be calculated as ∆ia = ua − va · dTs L (3.19) Since both ua and va are DC values, a constant k relating both could be used to simplify the equations. It expressed in terms as k= ua va (3.20) where k is a scalar value and the peak-peak current ripple expressed in terms of area EPS voltage is ∆ia = (k − 1) · dTs · va L (3.21) To solve for the average ISE inductor current, Equation 3.10 is used and the current is expressed as Ia = 1 − (2d − 1)k · va rL (3.22) 46 3.7. Initial Design of the ISE Since the ISE is designed to represent active or reactive elements, the ISE inductor current could also be expressed in terms of real and reactive power as √ 2 + Q2 PISE ISE = |Z ISE |−1 · va (3.23) Ia = va Equating Equation 3.22 to Equation 3.23, we can obtain a solution for the duty ratio in terms of P and Q as in d= (k + 1)|Z ISE | − rL 2k|Z ISE | (3.24) Having defined the duty cycle in terms of circuit parameters, the ripple ratio could be solved as in (k 2 − 1)Ts rL (k − 1)Ts ∆ia = · |Z ISE | − 2Ia 4kL 4kL = αISE [(k + 1)|Z ISE | − rL ] (3.25) (3.26) where αISE is a constant defined as αISE = (k − 1)Ts 4kL From the above equations, it is evident that the amount of ISE current ripple is related directly to the impedance being simulated within the ISE. Figure 3.12 illustrates how the DC gain factor k could be used as an extra degree of freedom to simulate ZISE at the same time limiting current ripple. For example, at a fixed ripple ratio of 10%, the ISE DC voltage source could be varied between k=1.1 to 1.8 to obtain a magnitude of simulated impedance in the range between 32 - 209 Ω. From Equation 3.26, it is noted that there is a minimum simulated ZISE at zero current ripple defined as ZISE = rL (k + 1) (3.27) 47 3.7. Initial Design of the ISE 400 300 z 200 100 0.20 0.15 0 0.10 1.2 1.4 k 0.05 1.6 1.8 ripple_ratio 0.00 Figure 3.12: Simulated impedance of ISE relationship to current ripple 48 Chapter 4 Improved Control Design for the ISE 4.1 Synchronous Frame Model of the Single-Phase ISE As in chapter 3, the phase equivalent circuits of the ISE is a single-phase PWM inverter as shown in Figure 4.1. The inverter is represented as a voltage source operating at fundamental frequency only and neglecting any higher switching frequency. In frequency domain, the ISE is modelled as a first-order system. The transfer function can be obtained in synchronous frame by selecting the inductor current as a state variable and the ISE generated voltage as the control input. The value of rL in Figure 4.1 represents the inductor resistance. Let the ISE DC bus voltage ua , inductor current ia and utility (area EPS) voltage va be expressed as va = Va · sin (ωt) Ia∗ · sin (ωt + ϕ) dia ua = rL · ia + ωL · + va dt ia = (4.1) (4.2) (4.3) where Va and Ia∗ are the area EPS voltage and reference current maximum amplitudes, respectively. In order to obtain the ISE voltage in synchronous Figure 4.1: Per-phase circuit diagram 49 4.1. Synchronous Frame Model of the Single-Phase ISE frame, a virtual second phase (B) that is 90o shifted from phase A is assumed and expressed as π vb∗ = Vb∗ · sin (ωt + ) (4.4) 2 π ib∗ = Ib∗ · sin (ωt + + ϕ) (4.5) 2 dib∗ ub∗ = rL · ib∗ + ωL · + vb∗ (4.6) dt Equation 4.3 and Equation 4.6 are combined in a two-phase system that is expressed as [ ] [ ] [ ] [ ] d ia ua ia va = rL · +L + (4.7) ub∗ ib∗ vb∗ dt ib∗ The input-to-output transfer function could be arrived at using the following two-phase Park’s transformation vdq = Sdq · vab∗ [ vd vq ] [ = cos (ωt) − sin (ωt) sin (ωt) cos (ωt) (4.8) ][ va vb∗ ] (4.9) For the above transformation into synchronous frame, the direct axis {d-axis} lags the quadrature axis {q-axis} by π/2. Since the dq-axis rotates at the system frequency counter-clockwise, values of the voltage and current transformed into this synchronous frame will have constant (DC) values. Multiplying Park’s transformation matrix, Sdq , by the system Equation 4.7 will yield ( [ ] ] [ ] [ [ ] [ ]) dSdq ia d ia ia va ua +L = rL · Sdq + Sdq + Sdq Sdq ib∗ ub∗ i i v dt dt b∗ b∗ b∗ (4.10) It can be seen that the differentiation of the transformation matrix with respect to time will introduce a cross-coupling voltage term between dq-axis and the synchronous frame system equation will be ] ] [ ] [ [ ] [ ] [ d id ud i −iq vd = rL · d + ωL · +L· + (4.11) uq iq id vq dt iq Taking the Laplace transform of Equation 4.11, the system equation in frequency domain will be [ ] [ ] [ ] [ ] Ud Id −Iq Vd = (sL + rL ) · + ωL · + (4.12) Uq Iq Id Vq 50 4.2. Synchronous Frame Model of the Three-Phase ISE Figure 4.2: Transfer function block diagram where Ud , Uq , Id , Iq , Vd and Vq are the Laplace transforms of the variable in Equation 4.11. The term (sL + rL ) is the inverse of the transfer function G(s) representing the ISE. The single-phase transfer function block diagram of the synchronous frame controlled ISE is shown in Figure 4.2. It shows that cross-coupling term between the d and q axis, ωL and −ωL, act as a feedback loop between them. 4.2 Synchronous Frame Model of the Three-Phase ISE Modelling the three-phase ISE is approached in the same manner as in the synchronous frame modelling of the single-phase ISE. From the per-phase three-phase PWM circuit of Figure 4.3, let the area EPS voltages, {va ,vb ,vc }, and the three-phase currents, {ia ,ib ,ic }, be expressed as 51 4.2. Synchronous Frame Model of the Three-Phase ISE Figure 4.3: Per-phase equivalent circuit of a three-phase PWM inverter va = Va · sin (ωt) 2π ) 3 2π vc = Vc · sin (ωt + ) 3 vb = Vb · sin (ωt − (4.13) ia = Ia∗ · sin (ωt + ϕ) 2π ib = Ib∗ · sin (ωt − + ϕ) (4.14) 3 2π ic = Ic∗ · sin (ωt + + ϕ) 3 where {Va ,Vb ,Vc } and {Ia∗ ,Ib∗ ,Ic∗ } are the maximum area EPS voltage and reference current amplitudes, respectively. For the equivalent model of Figure 4.3 and under balanced conditions, the circuit can be represented as a system of three equations as in ia ua va ia ub = rL · ib + L · d ib + vb (4.15) dt ic uc vc ic A transformation of the three-phase system of Equation 4.15 into synchronous reference frame is possible using Park’s transformation matrix Tdq0 52 4.2. Synchronous Frame Model of the Three-Phase ISE as in Tdq0 2π cos (ωt) cos (ωt + 2π 3 ) cos (ωt − 3 ) 2π = sin (ωt) sin (ωt + 2π 3 ) sin (ωt − 3 ) 1 2 1 2 where vdq0 could be obtained from vd va vq = 2 Tdq0 vb 3 v0 vc (4.16) 1 2 (4.17) The three-phase system of Equation 4.15 is transformed to synchronous reference frame using Tdq0 as follows ua ia ia ia va dTdq0 d ib +Tdq0 ib Tdq0 ub = rL·Tdq0 ib +L· +Tdq0 vb (4.18) dt dt uc ic ic ic vc And in the same manner as in Equation 4.10, udq0 is arrived at for the three-phase system and described as −iq vd id id ud uq = rL · iq +ωL· id +L· d iq + vq (4.19) dt i0 v0 i0 i0 u0 In frequency domain, the Laplace transform of Equation 4.19 is found to be Vd −Iq Id Ud Uq = (sL + rL )· Iq +ωL· Id + Vq (4.20) V0 I0 I0 U0 where Udq0 , Idq0 , and Vdq0 are the synchronous frame Laplace domain functions of the ISE voltages, ISE currents and area EPS voltages, respectively. The three-phase ISE transfer function in frequency domain is identical to Equation 3.14 and repeated here for convenience. G(s) = 1 rL + sL (4.21) It is worthy to note that by assuming a balanced three-phase system and neglecting the zero-sequence component of Equation 4.20, the system equation of the three-phase ISE is identical to that of the single-phase ISE in Equation 4.12. Hence the transfer function of the three-phase ISE is also represented in Figure 4.2. 53 4.3. Proposed Synchronous Frame Digital Current Control Figure 4.4: Proposed synchronous frame digital current control 4.3 Proposed Synchronous Frame Digital Current Control The proposed current control calculates a reference current command based on real and reactive power commands in synchronous frame. The input stage of the digital current controller of Figure 4.4 is the synchronous frame current commands. The second stage is a PI controller that will regulate the error between actual and command current signals. The decoupling stage outputs the ISE synchronous frame voltages Udq which are the input commands to the space vector PWM generator. The following sections will include detailed descriptions of the current reference calculation, synchronous frame current control and space vector PWM generation. 4.3.1 Current Reference Calculation The current command calculation in synchronous frame is expressed in terms of real and reactive power commands and inverter terminal voltages in synchronous frame according to ][ ∗ ] [ ∗ ] [ id vd −vq P = (4.22) ∗ iq vq vd Q∗ where {vd ,vq } are the ISE terminal voltages in synchronous reference frame and are equal to the area EPS voltages before island creation. {P ∗ ,Q∗ } are the command power desired and for unity power factor operation Q∗ is set to zero [22]. 54 4.3. Proposed Synchronous Frame Digital Current Control Figure 4.5: Closed loop current current control in synchronous frame 4.3.2 Synchronous Frame Current Control The closed loop synchronous frame current control block is detailed in Figure 4.5. The first stage is to compare the ISE currents in synchronous reference frame to the calculated reference currents according to the commanded currents {i∗d ,i∗q } from Equation 4.22 and then pass the error through a PI controller as expressed in ( ) 1 + sTi Gc (s) = Kp (4.23) sTi where Ki is defined as Kp /Ti . The second stage is used to remove the effect of the terminal voltages and the cross-coupling terms {-ωLIq , ωLId } in a forward path to obtain the ISE space-vector voltage commands {Ud∗ ,Uq∗ } in synchronous reference frame. It can be shown that the closed-loop transfer function for both d and q of Figure 4.5 is decoupled and simplifies to the block diagram of Figure 4.6. In order to realize the current control proposed, a discretized version of the closed-loop decoupled system is presented in Figure 4.7. The zero-order hold (ZOH) is used to represent the function of the pulse width modulation of the ISE. The transfer function of the ISE in discrete-time domain is 55 4.3. Proposed Synchronous Frame Digital Current Control Figure 4.6: Simplified reference frame current control transfer function Figure 4.7: Current control discrete-time block diagram represented as G(s) G(z) = (1 − z −1 ) Z( ) ( ) s 1 1−p = rL z−p (4.24) where p is defined in terms of the discrete time step Td as p = e−rL Td /L (4.25) The PI controller transfer function Gc (s) is transformed into discretetime domain and expressed as ( ) Kp z − zpi Gc (z) = (4.26) zpi z−1 56 4.3. Proposed Synchronous Frame Digital Current Control where zpi is defined as zpi = 1 1 + T1i (4.27) From Equation 4.24 and Equation 4.26 the open-loop transfer function in discrete-time domain is obtained as in ( ) Kp (1 − p) z − zpi Gc (z) · G(z) = (4.28) zpi rL z 2 − (1 + p)z + p For design of the current control system, complex z-domain closed-loop poles will be chosen based on performance specifications such as settling time and percent overshoot, Ts and P OS respectively. Let the complex closed-loop pole be defined as z ∗ = e(−σ±jωd )Td (4.29) where Td is the discrete time step and σ and the damped frequency ωd are defined as σ = ζ · ωn 4 ωn = ζ · Ts ωd = ωn √ 1 − ζ2 P OS = 100e n − πζω ω (4.30) d For pole placement of the closed-loop system, Figure 4.8 shows the uncompensated open-loop poles {p,1} and zero{z0 } in addition to the desired closed-loop pole {z ∗ }. Note that in order to satisfy angle condition, the closed-loop pole must satisfy the open-loop condition ∠{Gc (z)G(z)} = ±π(2k + 1) k = 0, 1, 2, . . . (4.31) From Figure 4.8, it can be shown that the PI controller zero could be found from zpi = ℜ{z ∗ } + h cos (π − βpi ) = ℜ{z ∗ } − h cos (βpi ) (4.32) 57 4.3. Proposed Synchronous Frame Digital Current Control Figure 4.8: Current control open-loop poles and zeros and desired pole z ∗ where h is expressed as h= ℑ{z ∗ } sin (βpi ) (4.33) Since Equation 4.31 must be satisfied, from the Figure β0 could be calculated as βpi = αp + α1 − π (4.34) based on βpi − αp − α1 = ±π(2k + 1) k = 0, 1, 2, . . . (4.35) while αp and α1 can be found from Figure 4.8 as αp = π − tan−1 α1 = π − tan−1 ( ( ℑ{z ∗ } p − ℜ{z ∗ } ℑ{z ∗ } 1 − ℜ{z ∗ } ) (4.36) ) (4.37) For the design of the PI controller, Kp and Ti are found from Equation 4.28 and the condition |Gc (z)G(z)|z=z ∗ = 1 (4.38) 58 4.3. Proposed Synchronous Frame Digital Current Control Root Locus 1 0.5π/T 0.6π/T 0.8 0.4π/T 0.1 0.3π/T 0.7π/T 0.2 0.3 0.6 0.8π/T 0.9π/T Imaginary Axis 0.2π/T 0.4 0.5 0.6 0.7 0.8 0.4 0.2 x 0.1π/T x 0.1π/T 0.9 π/T π/T 0 −0.2 0.9π/T −0.4 0.8π/T −0.6 0.2π/T Closed−Loop Poles 0.7π/T −0.8 0.3π/T 0.6π/T −1 −1 0.4π/T 0.5π/T −0.8 −0.6 −0.4 −0.2 0 Real Axis 0.2 0.4 0.6 0.8 1 (a) Root locus closed-loop poles of the current controller Step response of the closed loop synchrnous frame current control 1.4 1.2 Amplitude 1 0.8 0.6 0.4 0.2 0 0 0.001 0.002 0.003 0.004 0.005 Time (s) (sec) 0.006 0.007 0.008 0.009 0.01 (b) Step response of the current controller Figure 4.9: Design verification of the synchronous frame current controller to be rL · zpi (z − p)(z − 1) 1−p z − zpi zpi Ti = 1 − zpi Kp = (4.39) |z=z∗ (4.40) From system parameters and Equation 4.39 and Equation 4.40, the PI controller proportional constant Kp and integral time constant Ti can be found as in The new closed-loop pole placement is illustrated in Figure 4.9a. The new location of the designed complex conjugate poles {z ∗ } is calculated 59 4.4. Space Vector PWM (SVPWM) Generation Table 4.1: PI controller design parameters Parameter Symbol Value Inductor resistance rL 0.05 Ω Inductance L 10 mH Switching Frequency fd 10.8 kHz Settling time Ts 2 ms Percent overshoot P OS 4.3% Damping ratio ζ 0.707 Proportional constant Kp 14.28 Integral time-constant Ti 2.558 to be at 0.6174 ± j0.2610 which is within the unit circle and guaranteeing stability of the synchronous frame current controller. Also, it can be seen from the step response of the system in Figure 4.9b that the settling time and percent overshoot is as per design parameters. 4.4 Space Vector PWM (SVPWM) Generation The synchronous frame current controller will output the ISE dq command voltages in reference frame;{Vd∗ ,Vq∗ }. Once these space vector commands are generated, space vector pulse width modulation is implemented to obtain switch commands {SA , SB , SC } of Figure 4.5. To generate the switching commands, a symmetrical space vector representation of the ISE command voltage is generated using all possible switching states [23]. The choice to use SVPWM depended on its superb performance compared to conventional carrier-mode PWM. Also, lower switching loss, improved DC-link voltage utilization, and reduced switching ripple current are additional reasons why SVPWM was chosen for this design. Implementations of SVPWM for the three-phase ISE is not the focus of this thesis but it would have possible usage for future islanding tests or simple represent a three phase programable load capable of representing power functions. The application of SVPWM to control the ISE requires calculating the switching state ON-time duration. Two approaches to calculating the ONtime duration are available; using sampled reference space vector phase angle or amplitude. This design implements SVPWM using reference space vector phase angle to calculate ON-time duration. For reference, the derivation of ON-time duration using reference space vector phase angle and amplitude are included in Appendix (appendix svpwm.tex). 60 4.4. Space Vector PWM (SVPWM) Generation Q1 Q3 Q5 PVDG Vdc & VA VB VC DC/DC Q2 Converter Q4 Q6 Figure 4.10: Three-phase two-level inverter U~3 S2 U~2 S3 S1 t2 v~s∗ α U~4 t1 S4 U~1 S6 U~5 S5 U~6 Figure 4.11: Space vector representation for all possible switching states Figure 4.11 shows all possible switching states of the three-phase inverter ⃗ 1 to U ⃗ 6 have a magnitude of 2/3 Vdc ; shown in Figure 4.10. Active vectors U where Vdc is the dc-bus voltage, while states 000 and 111 correspond to the ⃗ 0 and U ⃗ 7 , respectively. S1 to S6 denotes sectors for all zero voltage vectors U ∗ active vectors. ⃗vs is defined as the synchronous frame reference space vector to be synthesized and is expressed as ⃗vs∗ = |vs∗ | [cos (ωt) + j sin (ωt)] (4.41) where ω is the system operating frequency. Table 4.2 shows the switching states and their corresponding space vectors. Once the sector of the reference space vector is identified, the process of synthesizing the reference space vector of Equation 4.41 using adjacent space vectors could be achieved using the ON-time duration Equation 4.42. 61 4.4. Space Vector PWM (SVPWM) Generation Table 4.2: Space vectors and corresponding switching states Space Vector Switching state U0 000 U1 100 U2 110 U3 010 U4 011 U5 001 U6 101 U7 111 ⃗vs∗ = ] 1 [ ti · Ui + t(i+1) · U(i+1) ts i = 1, 2, . . . , 6 (4.42) where ts is the switching time period and i is defined as the sector number7 . Solution for the ON-time duration for the adjacent active space vectors is included in Appendix A and Table 4.3 shows all sectors ON-time durations. Symmetrical implementation of SVPWM requires centering the active vectors in the first half of the switching period and padding the beginning and end of the half cycle with the zero vectors {U0 ,U7 } equally and insuring minimal switching sequence. The sequence pattern is then reversed for the other half of the cycle. A pattern of the switching sequence for all sectors is illustrated in Figure 4.12. PSIM was used to implement symmetrical SVPWM using a DLL file to generate the gate signals for the three-phase PWM inverter. A copy of the DLL file is attached to appendix (appendix svpwm.tex). The output voltages of the PWM inverter are shown in Figure 4.13. Here VA is the leg A voltage taken after the smoothing inductor with respect to the Y-connected load neutral point. While VAB is the phase-phase voltage and VGN is the voltage across the midpoint of the DC bus to the load neutral voltage. For demonstration, if there was a closed path for the triplen harmonic to pass to the midpoint of the DC voltage from the load neutral point, VA −VG +VN shows the flat top phase voltage that is the combination of the triplen harmonic and the fundamental voltage VA . 7 For sector 6, i+1 is reset to 1 to complete the synthesis cycle of Figure 4.11 62 4.4. Space Vector PWM (SVPWM) Generation Figure 4.12: Switching commands generated for SVPWM Figure 4.13: SVPWM phase A voltage outputs 63 4.5. Verification of Islanding PWM Inverter Performance Table 4.3: ON-time durations for three-phase SVPWM control Sector(i) 4.5 ABC 1 (100) 2 (110) 3 (010) 4 (011) 5 (001) 6 (101) ti t t(i+1) π |⃗vs∗ | sin 3 −α s ⃗ |U1 | sin π3 |⃗vs∗ | sin α s ⃗ sin π |U2 | 3 π |⃗vs∗ | sin 3 −α s ⃗ |U3 | sin π3 |⃗vs∗ | sin α s ⃗ sin π |U4 | 3 π |⃗vs∗ | sin 3 −α s ⃗ |U5 | sin π3 |⃗vs∗ | sin α s ⃗ sin π |U6 | 3 ( t t ( ) ( ) ( ) t t ) ( ) ( ) ( ( ) t ( ) ) ∗ ts |⃗v⃗s | sinsin απ |U2 | (3) π |⃗vs∗ | sin ( 3 −α) ts ⃗ sin π |U3 | (3) ∗ ts |⃗v⃗s | sinsin απ |U4 | (3) π |⃗vs∗ | sin (α− 3 ) ts ⃗ sin π |U5 | (3) ∗ ts |⃗v⃗s | sinsin απ |U6 | (3) π |⃗vs∗ | sin ( 3 −α) ts ⃗ sin π |U1 | (3) Range 0≤α< π 3 ≤α< 2π 3 π 3 2π 3 ≤α<π −π ≤ α < − 2π 3 π − 2π 3 ≤ α < −3 − π3 ≤ α < 0 Verification of Islanding PWM Inverter Performance To verify the proper performance of the inverter designs, an islanding situation was created using the unintentional islanding test setup and the inverter was allowed to continue to operate under islanding conditions. This is to insure a proper test bench for further development of islanding detection schemes and provide a comparison case for the proposed improvements to the islanding test. 4.5.1 Verification of Single-Phase PWM Inverter Islanding Figure 4.14 shows a single phase inverter under islanding condition. The PWM inverter continues normal operation after utility disconnection. There is almost no change in amplitude since the real power balance was achieved before disconnection. The frequency of the inverter also remains within nominal operating limits {59.3 - 60.5 Hz}. From Figure 4.14a the output current is in phase with the terminal voltage indicating unity power factor operation. From Figure 4.14b the inverter output real and reactive power are shown. Here the inverter starts operation at 0.03 sec and the utility real power delivery is seen to reduce as the inverter real power starts supplying the simulated island load. At time 0.15 sec, utility disconnection occur and the reactive power is seen to adjust to almost zero. The small change in reactive power mismatch does not force any parameter outside threshold limits and the inverter continues operating under islanding conditions. 64 4.5. Verification of Islanding PWM Inverter Performance (a) Voltage, current, and frequency (b) Active and reactive power Figure 4.14: Single-phase inverter output under islanding condition 65 4.5. Verification of Islanding PWM Inverter Performance 4.5.2 Verification of Three-Phase PWM Inverter Islanding The same procedure and simulated island parameters were used to test a three-phase PWM inverter for islanding. Figure 4.15a shows the output voltages, currents and phase A voltage and current frequencies. The frequency and amplitude of both voltage and current of the inverter remain with in nominal parameters after utility disconnection at time 0.25 sec. Figure 4.15b shows phase A real and reactive power outputs and as expected, almost no power imbalance present. 4.5.3 Islanding Detection Method Employed From the review of islanding detection methods in the previous chapter, many islanding techniques could be implemented into the control of the PWM inverter to prevent islanding. Since the focus of this work is not the islanding detection scheme but rather the test setup itself, the islanding scheme is chosen based on simplicity of implementation. Active frequency drift (AFD) was chosen to be implemented in the single-phase PWM inverter control. Figure 4.16 shows the design control used to drift the frequency of the reference current. The drift frequency (∆f ) was set to 5% of the nominal operating frequency. To generate the reference signal, the voltage at the point of common coupling (Vpcc ) is sensed and the period is extracted and fed as an input to the control of the pwm inverter. Frequency drift is generated from the equation i∗ (t) = P∗ sin [2π(fo ± ∆f )t + ϕ] Vm (4.43) where i∗ (t) is the instantaneous reference command current, P ∗ is the command real power, Vm is the peak nominal voltage, fo is the nominal frequency, ∆f is the drift in frequency and ϕ is the voltage phase angle. The single-phase inverter output voltage and current are shown in Figure 4.17. It can be seen that the drift in frequency after utility disconnects at 0.15s and a grid-tie disconnection signal is activated 26.6ms after disconnection. Negative sequence injection was chosen to be implemented in the three-phase PWM inverter control. This case is based on generating an unbalanced reference current commands for the space vector modulation of 66 4.5. Verification of Islanding PWM Inverter Performance (a) Voltage, current, and frequency (b) Active and reactive power Figure 4.15: Three-phase inverter output under islanding condition 67 4.5. Verification of Islanding PWM Inverter Performance Figure 4.16: AFD implemented in the single-phase PWM control Figure 4.17: AFD implemented in the single-phase PWM control 68 Negative Seq. (V rms) 4.5. Verification of Islanding PWM Inverter Performance 10 0% inject. − avg=1.78 8 6 4 2 0 Negative Seq. (V rms) 0.5 1 1.5 2 2.5 10 2% inject. − avg=2.36 8 6 4 2 0 0.5 1 1.5 2 2.5 Negative Seq. (V rms) 10 5%n inject. − avg=2.36 8 6 4 2 0 0.5 1 1.5 2 2.5 Time (s) Figure 4.18: Negative sequence voltage injection at 0, 2 and 5% Figure 4.4. The synchronous frame Vd∗ and Vq∗ of the reference current control are transformed into rotating frame and injected with a set of three 5% perturbation shifted by 120o at each phase zero crossing. Then the modified rotating reference current commands transformed back to synchronous frame for space vector modulation processing to generate the modified gate signal commands for the three-phase inverter. Figure 4.18 shows three cases of negative sequence injection 0, 2 and 5% of the maximum nominal voltage. Also, Figure 4.19 shows the frequency response for the same cases. It can be seen that at the moment of disconnection there is a sharp rise in negative sequence content compared with the no injection case especially at 5% negative sequence injection where the frequency is pushed out of normal operating boundary limit of 60.5Hz. Detection of islanding occur within the first cycle after disconnection using the modified SVPWM with 5% negative sequence injection. Figure 4.20 shows the 5% negative sequence frequency of the voltage at PCC. The frequency reaches the upped frequency limit at close to half a cycle and confirmation of islanding is complete within a cycle of Area EPS disconnection. 69 4.5. Verification of Islanding PWM Inverter Performance 61 Frequency (Hz) 0% inject. 60.5 60 59.5 59 0.5 1 1.5 2 2.5 61 Frequency (Hz) 2% inject. 60.5 60 59.5 59 0.5 1 1.5 2 2.5 61 Frequency (Hz) 5% inject. 60.5 60 59.5 59 0.5 1 1.5 2 2.5 Time (s) Figure 4.19: 0%, 2% and 5% negative sequence injection 60.7 Detection confirmed within one cycle @ 5% negative Sequence Injection 60.6 60.5 Frequency (Hz) 60.4 Frequency signal for reference only 60.3 60.2 Area EPS Disc 60.1 60 59.9 0.24 0.245 0.25 0.255 0.26 0.265 0.27 0.275 Time (s) Figure 4.20: Output of negative sequence islanding detection 70 Chapter 5 Modified Unintentional Islanding Test 5.1 Introduction In this chapter, the modified islanding test is presented and experimental results are compared to the results of the standard test. The first case presented is a verification of the ISE’s ability to establish a power island as in case 1 of the standard islanding test in chapter 2. The next two cases will test two inverters one with AFD islanding detection and the other with RPV detection scheme. These cases match those used in chapter 2. In addition, there are two more cases testing three-phase grid-connected inverts one without islanding detection and the other with negative sequence injection for islanding detection. The modified islanding test circuit is presented in Figure 5.1. The EUT is highlighted in green while the test island is within the yellow shade. The island includes a shedding resistor in parallel with the ISE. It is noted here that this setup is fixed and the only physical connections needed are the two terminals of the single phase EUT. The circuit of Figure 5.1 is setup to test a single or three phase PWM inverter identical to that used in the standard test of chapter 2. The ISE could be set to compensate for any mismatch in P, Q or both using PCT P and PCT Q controls. These parameters could be set as a percentage of compensation. To maintain minimum power requirement for the ISE, the P CMD and Q CMD are calculated based on the EUT real and reactive power measured. This will maintain near zero power from or to the area EPS. L SELECT is the control that will specify the mismatch in reactive power between L and C; it represents the quality factor Qf . L SELECT also represent the quality factor desired. 71 5.2. Case 1: A Single-Phase Islanding Inverter Figure 5.1: Modified island load with ISE 5.2 Case 1: A Single-Phase Islanding Inverter The islanding detection function has been disabled in the single phase inverter under test in Figure 5.1. The ISE was set with initial parameters as in Table 5.1. Table 5.1: ISE case 1 parameters Parameter Single-phase Pload Qload Vpcc Iinv fo fs R P Finv L SELECT 1 0 120 8.33 60 10 14.4 1 100 kW kVAR Vrms Arms Hz kHz Ω % The purpose for testing an islanding inverter it to confirm the ability to create an island. The inverter was tested in chapter 2 and results of this test will be presented and compared to the standard test of the islanding inverter. 72 5.2. Case 1: A Single-Phase Islanding Inverter 5.2.1 Inverter Voltage & Frequency As can be seen from Figure 5.2 and Figure 5.3 at t = 1sec the grid disconnects and there is hardly any noticeable change in the magnitude of VP CC or frequency. Figure 5.2 shows the voltage and current of the inverter only at zero mismatch (Qf =% 100). VPCC RLC indicates the voltage of the point of common coupling for the standard test. The rest of the curves refer to the different options the ISE can be set to. (100 0 0) means the ISE does not compensate for P and Q of the EUT. While (100 0 1) means the ISE compensate for Q of the EUT in addition to the Qload . (100 1 0) means that the ISE will only compensate for PEU T in addition to Pload . It is noted here that non of the ISE compensation has significant affect on V or I of the inverter since there is no islanding detection method employed nor there is mismatch in the load simulated LC represented within the ISE. Figure 5.2: VPCC and IEU T under zero mismatch condition8 . For an islanding inverter voltage and frequency are expected to remain within allowed operating limits. In the case of a exact mismatch, i.e. ∆P and ∆Q is zero, the voltage at PCC and frequency of the inverter should not exhibit any changes. Figure 5.3 shows the voltage magnitude, frequency, 8 Variables with RLC refer to the standard test and values within parenthesis refer to the modified test. The values refer to L SELECT, PCT P and PCT Q respectively. 73 5.2. Case 1: A Single-Phase Islanding Inverter 61.5 0 0.5 1 60.5 60 Grid Disconnect Frequency (Hz) 61 Grid Disconnect Voltage magnitude (V) Islanding test for inverter with AFD at Qf=1+/-0.05 190 185 180 175 170 165 160 155 150 145 59.5 59 58.5 1.5 2 Time (sec) 2.5 3 3.5 0 0.5 1 1.5 2 Time (sec) 2.5 3 3.5 1.5 0.5 0 Grid Disconnect Current (A) 1 -0.5 -1 -1.5 0.96 0.98 1 Time (sec) VPCC IINV 1.02 1.04 Figure 5.3: Frequency and voltage and inverter current results under islanding condition inverter current and voltage at PCC for the modified unintentional islanding test. The voltage magnitude for the all increments of the quality factor, Qf , shown remain almost unchanged. The frequency of the inverter shown in Figure 5.3 indicate that, even with the deliberate power mismatch, remain within the operational limits after grid disconnection. Finally, the inverter current and voltage are shown in the figure for Qf = 1. After grid disconnection, the inverter maintains the island as seen after t = 1sec. Here, the inverter current remains sinusoidal and the voltage at PCC follows the current in the absence of the EPS source. 5.2.2 Inverter Real and Reactive Power The real and reactive power results of the modified islanding test shown in Figure 5.4. The real power in the figure is maintained at the rated value before and after grid disconnection. While the reactive power of the inverter is zero before the grid disconnection and any power mismatch in the reactive 74 5.3. Case 2: A Single-Phase Inverter with AFD 100 Power (VAR) 1020 1000 Grid Disconnect Power (W) 1040 980 960 0 0.5 1 50 0 Grid Disconnect Islanding test for inverter with AFD at Qf=1+/-0.05 -50 -100 1.5 2 Time (sec) 2.5 3 3.5 0 0.5 0 0.5 1 1.5 2 Time (sec) 2.5 3 3.5 1.5 2 Time (sec) 2.5 3 3.5 100 0 -20 -40 0 0.5 1 50 0 Grid Disconnect EPS Power (VAR) 20 Grid Disconnect EPS Power (W) 40 -50 -100 1.5 2 Time (sec) 2.5 3 3.5 1 Figure 5.4: P and Q of inverter results under islanding condition power is supplied by the grid. After the grid disconnection, the reactive power has to be sourced by the inverter which will cause the frequency drift to a new operating point for power matching. Figure 5.4 also shows the real and reactive power of the EPS. PEP S remains at almost zero since real power is matched, while the EPS provided the reactive power, QEP S , required due to the mismatch in reactive power due to the incremental changes in the power quality factor. 5.3 Case 2: A Single-Phase Inverter with AFD The setup shows the test performed on an islanding inverter with active frequency islanding detection. The islanding detection method applies 5% crescent factor to the inverter frequency every half cycle. While the inverter is connected to the grid, the frequency is maintained at grid frequency and the current would remain at zero level until the next voltage zero crossing of the grid (Vpcc). In Figure 5.5, voltage , frequency and current of the inverter is shown. 75 5.3. Case 2: A Single-Phase Inverter with AFD After grid disconnection, the voltage is maintained as the ISE supplies the reactive power needed to maintain power balance and virtual zero mismatch. 61.5 190 185 180 175 170 165 160 155 150 145 0 0.5 1 60.5 60 Grid Disconnect Frequency (Hz) 61 Grid Disconnect Voltage magnitude (V) Islanding test for inverter with AFD at Qf=1+/-0.05 59.5 59 58.5 1.5 2 Time (sec) 2.5 3 3.5 0 0.5 1 1.5 2 Time (sec) 2.5 3 3.5 1.5 0.5 0 Grid Disconnect Current (A) 1 -0.5 -1 -1.5 0.96 0.98 1 Time (sec) VPCC IINV 1.02 1.04 Figure 5.5: Voltage and frequency response of inverter with AFD The amplitude of the voltage at the PCC remains almost the same while the frequency shifts higher outside the upper operating limit. Here, the inverter internal over frequency protection will engage and an inverter grid disconnection signal is initiated. The current of the inverter before and after disconnection is maintained except for a frequency shift that occurs at t=1 sec. It is noted here that the figure Figure 5.5 shows the results of the test at quality factor of 1 with real and reactive power compensation engaged in the ISE. Comparing these results with the result of the AFD test performed using the standard test would reveal in Figure 5.6 that the ISE test would fail the inverter for the quality factor range of 0.95, 0.96, and 0.97. When the ISE test composition for the real power would increase the reaction of the inverter to push the frequency outside the operating limits as in figure Figure 5.7. While the real power of the inverter is constant, the ISE can supply any mismatch in real power due to the islanding detection method used in the inverter, thus, increasing the accuracy of the test and 76 5.3. Case 2: A Single-Phase Inverter with AFD 100 Power (VAR) 1020 1000 Grid Disconnect Power (W) 1040 980 960 0 0.5 1 50 0 Grid Disconnect Islanding test for inverter with AFD at Qf=1+/-0.05 -50 -100 1.5 2 Time (sec) 2.5 3 3.5 0 0.5 0 0.5 1 1.5 2 Time (sec) 2.5 3 3.5 1.5 2 Time (sec) 2.5 3 3.5 100 0 -20 -40 0 0.5 1 50 0 Grid Disconnect EPS Power (VAR) 20 Grid Disconnect EPS Power (W) 40 -50 -100 1.5 2 Time (sec) 2.5 3 3.5 1 Figure 5.6: Real and reactive power of the inverter with AFD 77 5.4. Case 3: A Single-Phase Inverter with RPV providing zero mismatch conditions. Frequency: Standard Test vs Modified Test 61.5 Black: Modified Test Red: Standard Test 61 fmax Frequency (Hz) 60.5 60 59.5 Grid Disconnect fmin 59 58.5 0 0.5 1 1.5 2 Time (sec) 2.5 3 3.5 Figure 5.7: Frequency results comparison between the standard test and the modified test A comparison of the results are presented in figure Figure 5.8 between the standard islanding test and the modified proposed test. Here the ISE acting as a compensation device, ensuring that the islanding condition is maintained. The modified test results closely matches that of the standard test. It is noted here that the ISE frequency result is for reactive power compensation from the ISE which monitors the inverter reactive power. It is noted here that Figure 5.8 shows the influence of reactive power compensation due to the reactive power produced by the inverter due to implementing the active frequency drift islanding detection method. In this case, the reactive power in this system will no longer match that of the standard test. It also will push the inverter outside the frequency operational limit for all increments of the quality factor. 5.4 Case 3: A Single-Phase Inverter with RPV In this third case, the inverter islanding detection method is the reactive power variation. Here, the inverter is varying the inverter’s reactive power over 1 second with 5% of the rated power. Figure 5.9 shows the results of voltage, frequency and inverter current. Again, the voltage magnitude remains within operational limits as expected. While the frequency is maintained at 60Hz before grid disconnection, the 78 5.4. Case 3: A Single-Phase Inverter with RPV Frequency: Standard Test vs Modified Test 61.5 Black: Modified Test Red: Standard Test 61 fmax Frequency (Hz) 60.5 60 59.5 Grid Disconnect fmin 59 58.5 0 0.5 1 1.5 2 Time (sec) 2.5 3 3.5 Figure 5.8: Frequency results comparison with reactive power compensation frequency varies after grid disconnects due to the variation of reactive power. Figure 5.10 shows the real and reactive power of the inverter and EPS. The inverter active power is maintained at the rated value as expected even though its detection method is trying to vary the reactive power periodically. It is noticed in the reactive power graph that before grid disconnection the mismatch in power is supplied by the EPS, while after grid disconnection the reactive power remains constant as specified in the unintentional islanding test. Figure 5.11 shows the standard test frequency results of chapter 2 in comparison to the modified test using the ISE. It is shown that both test results closely match which implies that the proposed active method is adequate for improving on the test’s portability, automation and cost effectiveness. Finally, when enabling the reactive power compensation of the ISE, it is shown that the range of frequency outside the operational limit is smaller due to the smaller mismatch in reactive power. Hence, the non-detection zone is larger than that of the standard test and a question arises if this is a situation that could relate to multiple penetration case of inverters and islanding. It is suggested here to further investigate this case as part of future work. 79 5.4. Case 3: A Single-Phase Inverter with RPV 61.5 0 0.5 1 60.5 60 Grid Disconnect Frequency (Hz) 61 Grid Disconnect Voltage magnitude (V) Islanding test for inverter with AFD at Qf=1+/-0.05 190 185 180 175 170 165 160 155 150 145 59.5 59 58.5 1.5 2 2.5 3 3.5 0 0.5 1 Time (sec) 1.5 2 2.5 3 3.5 Time (sec) 1.5 0.5 0 Grid Disconnect Current (A) 1 -0.5 -1 -1.5 0.96 0.98 1 Time (sec) VPCC IINV 1.02 1.04 Figure 5.9: Voltage and frequency response of inverter with RPV 80 5.4. Case 3: A Single-Phase Inverter with RPV 100 1000 980 960 0 0.5 1 50 Power (VAR) 1020 Grid Disconnect Power (W) 1040 0 Grid Disconnect Islanding test for inverter with AFD at Qf=1+/-0.05 -50 -100 1.5 2 Time (sec) 2.5 3 3.5 0 0.5 0 0.5 1 1.5 2 Time (sec) 2.5 3 3.5 1.5 2 Time (sec) 2.5 3 3.5 150 0 -20 -40 0 0.5 100 50 0 Grid Disconnect EPS Power (VAR) 20 Grid Disconnect EPS Power (W) 40 -50 -100 -150 1 1.5 2 Time (sec) 2.5 3 3.5 1 Figure 5.10: Real and reactive power of the inverter with RPV Frequency: Standard Test vs Modified Test 61.5 Black: Modified Test Red: Standard Test 61 fmax Frequency (Hz) 60.5 60 59.5 Grid Disconnect fmin 59 58.5 0 0.5 1 1.5 2 Time (sec) 2.5 3 3.5 Figure 5.11: Frequency results comparison between the standard test and the modified test 81 5.4. Case 3: A Single-Phase Inverter with RPV Frequency: Standard Test vs Modified Test 61.5 Black: Modified Test Red: Standard Test 61 fmax Frequency (Hz) 60.5 60 59.5 Grid Disconnect fmin 59 58.5 0 0.5 1 1.5 2 Time (sec) 2.5 3 3.5 Figure 5.12: Frequency results comparison between the standard test and the modified test with reactive power compensation 82 Chapter 6 Experimental Development 6.1 Hardware Circuit A prototype ISE was built and developed and the circuit built is shown in Figure 6.5. The hardware in the figure consists of: • A control board that includes sensors, DSP, and power supplies is shown in Figure 6.3 • The IGBT driver circuit and IGBT H-bridge inverter is shown in Figure 6.4. • A DC power supply (HPD 30-10 by Xantrex) which can supply 60V and 3A DC. • A 3-Φ AC source and a 3-Φ Transformer. • Two series inductors (2.6mH and 26mH) for ripple study. The circuit was assembled was connected with the grid using a variable voltage transformer. The ISE in this setup was tested to represent an inductor, a capacitor and a resistor. This will ensure the proper functioning of the ISE in the islanding test. 6.2 Simulated Inductance (Lsim ) The ISE simulated the mismatch in LC in addition to supplying reactive power, if commanded, to compensate for any generated Q by the EUT. From Equation 3.25, the parameters for different inductance ranges can be found and shown in Table 6.1. In the table, two values of series inductance, 2.6mH and 26mH, are evaluated at a switching frequency of 20kHz and a voltage ratio k = VDC /Vg = 1.5. 83 6.3. Simulated Capacitance (Csim ) Lsim (Csim ) 0.5 − 1p.u. 1 − 10p.u. 10 − 100p.u. Inductor L 2.6mH 26mH 260mH Resolution (%) ≤ 0.00089 ≤ 0.090 ≤ 0.90 Current Ripple ≤ 0.24 ≤ 0.24 ≤ 0.24 Table 6.1: ISE inductive or capacitive load specification under fs = 20kHz, k = 1.3, rL = 0.5Ω and the impedance base value ∥Z∥ = 40Ω For series inductance of 2.6mH, the ISE output was obtained as in Figure 6.6 for a inductance of 0.5pu and in Figure 6.7 the simulated inductance was 1pu. In Figure 6.6 and Figure 6.7, VLsim represents the voltage across the ISE which is also the voltage VP CC . From the figures, the magnitude of VLsim is approximately Vg but the phase angle is shifted by a 90o phase delay with respect to Vg . It is noted here that the current ripple, as predicted in chapter 3, increases with the impedance of Lsim . From the figure, with the increase of the series inductance from 2.6mH to 26mH the current ripple decreases dramatically to 0.1 of the current ripple of the smaller series inductance. 6.3 Simulated Capacitance (Csim ) The ISE representing a capacitor load only requires the phase angle which leads the source voltage Vg by 90o . Figure 6.8 and Figure 6.9 shows the ISE representing a capacitance with a range as deduced in chapter 3. Here the figure shows two per-unit values of the capacitance which is sufficient for use in the islanding test to represent the mismatch in the load as specified in the unintentional islanding test. 6.4 ISE Series Inductance and Switching Frequency To manage the ripple according to Equation 3.21, the series inductance of the ISE or the chosen value of voltage ratio or both can be chosen to improve the current ripple. For comparison, two values of series inductance, 2.6mH and 26mH, were used in the experiment to show current ripple reduction. Figure 6.14 84 6.4. ISE Series Inductance and Switching Frequency to Figure 6.17 shows the voltage output of the simulated inductance (VLsim ) and the source voltage (Vg )for comparison. 6.4.1 ISE Current Ripple Experimental Results and Analysis Series inductance of the ISE is inversely proportional to the current ripple as per Equation 3.21 and the results from the experiment are shown in Figure 6.10 to Figure 6.13. Two ISE series inductors, 2.6mH and 26mH, are chosen and a switching frequency is set to 20kHz. The current ripple using Lseries = 2.6mH is approximately 8 times of the current ripple. When using a series inductor of value 26mH, the output current has 20% error compared with the theoretical value “10 times”. Taking into account that the source voltage of the EPS had large harmonic noise which affect the errors. 6.4.2 Lsim ISE Current Ripple As the calculated current ripple in chapter 3 is defined, it should be reduced within acceptable range. Here, two series inductances are used separately to verify the reduction of the current ripple as shown in Figure 6.14 to Figure 6.17. 6.4.3 Csim ISE Current Ripple Current ripple is demonstrated in Figure 6.18 to Figure 6.21. It can be seen the big effect of the series inductance on the amount of ripple especially when the source includes high-frequency noise that is not desirable. 85 6.4. ISE Series Inductance and Switching Frequency Figure 6.1: Experimental system Figure 6.2: board Control circuit Figure 6.3: board Figure 6.4: The whole circuit Power circuit Figure 6.5: ISE hardware prototype system 86 6.4. ISE Series Inductance and Switching Frequency Figure 6.6: Lsim = 0.5p.u. at fpwm = 20kHz Figure 6.7: Lsim = 1p.u. at fpwm = 20kHz 87 6.4. ISE Series Inductance and Switching Frequency Figure 6.8: Csim = 0.5p.u. at fpwm = 20kHz Figure 6.9: Csim = 1p.u. at fpwm = 20kHz 88 6.4. ISE Series Inductance and Switching Frequency VRsim and Vg (Rsim = 2pu & L = 2.6mH) VRsim and Vg (Rsim = 2pu & L = 2.6mH) -5 VRsim (V) VRsim (V) 10 0 -10 -0.1 -10 ΔiL/2IL=0.46 -15 -0.08 -0.06 -0.04 -0.02 -0.0354 0 -0.0352 -0.035 -0.0348 -0.0346 -0.0344 -0.0352 -0.035 -0.0348 -0.0346 -0.0344 -6 Vg (V) Vg (V) 10 0 -10 -0.1 -8 -10 -12 -0.0354 -0.08 -0.06 -0.04 -0.02 ΔVg/Vg =0.22 0 Time (seconds) Figure 6.10: L = 2.6mH, Rsim = 2p.u. and fpwm = 20kHz Figure 6.11: Current Ripple at L = 2.6mH VRsim and Vg (Rsim = 2pu & L = 26mH) -6 VRsim (V) VRsim (V) 10 5 0 -5 -10 -0.1 -0.08 -0.06 -0.04 -0.02 -8 ΔiL/2IL=0.06 -0.0478 -0.0476 -0.0474 -0.0472 -0.047 -5 VRsim (V) VRsim (V) VRsim and Vg (Rsim = 2pu & L = 26mH) -7 -9 -0.048 0 10 5 0 -5 -10 -0.1 Time (seconds) -6 -7 -8 -0.08 -0.06 -0.04 -0.02 0 Time (seconds) Figure 6.12: L = 26mH, Rsim = 2p.u. and fpwm = 20kHz ΔVg/Vg =0.18 -9 -10 -0.048 -0.0478 -0.0476 -0.0474 Time (seconds) -0.0472 -0.047 Figure 6.13: Current ripple at L = 26mH 89 6.4. ISE Series Inductance and Switching Frequency Figure 6.14: L = 2.6mH ,Lsim = 1p.u. and fpwm = 20kHz Figure 6.15: L = 26mH ,Lsim = 1p.u. and fpwm = 20kHz VLsim Vg 1pu) VLsim and Vgand (26mH VLsim Vg (1pu) VLsim and and Vg (2.6mH 1pu) 4.7 Voltage (V) Voltage (V) 6 ΔiL/2IL=0.5 5 4 3 ΔiL/2IL=0.05 4.5 Vg 4.4 4.3 4.2 2 -0.0885 VLsim 4.6 -0.0885 -0.0884 -0.0884 Time (seconds) -0.0883 Figure 6.16: Current ripple at L = 2.6mH -0.0954 -0.0953 -0.0953 Time (seconds) -0.0952 Figure 6.17: Current ripple at L = 26mH 90 6.4. ISE Series Inductance and Switching Frequency Figure 6.18: L = 2.6mH ,Csim = 1p.u. and fpwm = 20kHz Figure 6.19: L = 26mH ,Csim = 1p.u. and fpwm = 20kHz VCsim Vg (1pu)1pu) VLsim andand Vg (2.6mH 4.8 Vg ΔiL/2IL=0.95 3 Voltage (V) Voltage (V) andVV (26mH1pu) 1pu) VVLsim and (2.6mH gg Csim VCsim 4 2 1 VCsim 4.6 Vg ΔiL/2IL=0.1 4.4 4.2 4 -0.0397 -0.0397 Time (seconds) -0.0396 Figure 6.20: Current ripple at L = 2.6mH -0.0774 -0.0774 -0.0773 -0.0772 Time (seconds) -0.0772 Figure 6.21: Current ripple at L = 26mH 91 Chapter 7 Conclusion 7.1 Improvements to the Standard Test The standard unintentional islanding test was examined, analyzed, and studied in depth. The test established a standard for insuring that any distributed generation system meant for connection to the grid has a proper function preventing it from creating and sustaining an unintentional island. The test uses discrete electronic elements, namely resistance, inductance and capacitance to simulate a grid island load. With the progress and development of technology, the rating of DG units intended for grid connection increased. This meant that larger island load is needed to perform the test. The discrete island load used in the test would become larger to meet DG rating requirement, which means higher cost and longer setup time. In addition, the portability and reproducibility of the test becomes more difficult as power rating increase. Looking at the history of the development of the unintentional islanding test, it was evident that the simplest form of the test is when the load was represented as a simple resistance. This meant that the real power is matched with the DG unit under test. Real network situation necessitate the presence of a reactive element to properly represent the EPS network. Although the LC elements provided resonance at utility frequency and any mismatch in reactive power could be represented by calculating the proper values of L and C, the bulk reactive power remained within the LC elements and only mismatch by design was evident in the circuit. Starting from this concept, the idea of this thesis started and through research, analysis and design. This research contribution could be summarized in the following: • The island stabilizing element (ISE) was designed and developed to function as the source of the mismatch in reactive power and replaced the LC elements in the standard test. • The stabilizing element was compact, cost effective and simple to adjust and move when compared to the standard test’s discrete static 92 7.2. Limitations of Proposed Test Setup LC load. • The ISE application scope is not limited to the unintentional islanding test, but also could function as an electronic load or a differential VAR compensator. • The design challenge to track Area EPS and EUT power flow and generate required reactive power was over come using a combination of control strategies. Mainly, Synchronous frame control and space vector PWM for optimal switching. • The performance of the ISE to reproduce the standard islanding test without the LC elements gave accurate results within the range of the simulated island impedance. 7.2 Limitations of Proposed Test Setup Chapter 3 showed in depth analysis of the limitation of the island stabilizing element in terms of simulated impedance that could be represented. Current ripple and range of impedance were some of the criteria that limited the application of the stabilizing element. With proper design, These limitations could be controlled according to the application and meeting power demand and ratings including limiting unwanted current ripple that could affect the results of the islanding test. Another limitation of the proposed test has to do with the resistor. Even though the stabilizing element could represent the function of the LC elements, more work needs to be done to represent the total island in a way that would save energy otherwise wasted as heat. Proper choice of DC to AC voltage ratio could improve on range of simulated impedance in the stabilizing element as mentioned in chapter 3 in addition to proper choice of the series inductor of the stabilizing element inverter. 7.3 Broader Application and Future Work The stabilizing element is basically an AC electronic load that is designed to purely represent reactive components as shown in the experimental results in chapter 6. Therefore, the application of the stabilizing element could be for AC circuits with tuned reactive elements as needed. 93 7.3. Broader Application and Future Work Further development of the stabilizing element could make it a full functional AC electronic load that is not limited to reactive power. Also, the design of the stabilizing element could be further improved as a multilevel current controlled current source capable of representing a larger rangle of AC/DC electronic loads. A more robust island stabilizing element with a wider impedance range is required to meet the demand of current and future DG units meant for testing. Improvement of the impedance range could be achieved with a multilevel inverter or parallel operation of a number of ISE units synched. In addition, filter design to eliminate current ripple is needed to improve the performance and quality of the ISE. 94 Bibliography [1] IEEEStandards, “Ieee standard for interconnecting distributed resources with electric power systems,” IEEE Std 1547-2003, vol. 1, p. 0, 2003. [2] I. Standards, “Ieee recommended practice for utility interface of photovoltaic (pv) systems,” IEEE Std 929-2000, vol. 1, pp. –, 2000. [3] B. Verhoeven, “Probability of islanding in utility networks due to grid-connected photovoltaic power systems,” KEMA Nederland B.V., Utrechtseweg 310, P.O.box 9035 6800 ET, Arnhem, The Netherlands, Tech. Rep. IEA PVPS T5-07: 2002, September 2002. [Online]. Available: http://www.iea-pvps.org/products/rep5 07.htm [4] B. Singam and L. Hui, “Assessing sms and pjd schemes of anti-islanding with varying quality factor,” Power and Energy Conference, 2006. PECon ’06. IEEE International, pp. 196–201, Nov. 2006. [5] S.-I. Jang and K.-H. Kim, “Development of a logical rule-based islanding detection method for distributed resources,” Power Engineering Society Winter Meeting, 2002. IEEE, vol. 2, pp. 800–806 vol.2, 2002. [6] M. Ropp, “Design issues for grid-connected photovoltaic systems,” Ph.D., Georgia Institute of Technology, Atlanta, GA, 1998. [7] G. Kern, “Sunsine300, utility interactive ac module anti-islanding test results,” Photovoltaic Specialists Conference, 1997., Conference Record of the Twenty-Sixth IEEE, pp. 1265–1268, Sep-3 Oct 1997. [8] M. Bower, W. Ropp, “Evaluation of islanding detection methods for photovoltaic utility-interactive power systems,” Sandia National Laboratories Photovoltaic Systems Research and Development, Albuquerque, NM, 87185-0753, USA (1) South Dakota University, Electrical Engineering Department HH205, Brookings, SD 570072220, USA (2), Report IEA PVPS T5-09: 2002, March 2002, 95 Bibliography 57 pages. [Online]. Available: download/rep5 09.pdf http://www.iea-pvps.org/products/ [9] G.-K. Hung, C.-C. Chang, and C.-L. Chen, “Automatic phase-shift method for islanding detection of grid-connected photovoltaic inverters,” Energy conversion, ieee transactions on, vol. 18, no. 1, pp. 169– 173, Mar 2003. [10] P. Mahat, Z. Chen, and B. Bak-Jensen, “Review of islanding detection methods for distributed generation,” in Proc. Third International Conference on Electric Utility Deregulation and Restructuring and Power Technologies DRPT 2008, 2008, pp. 2743–2748. [11] J. Stevens, S. Bonn, R.H., J. Ginn, S. Gonzalez, and G. Kern, “Development and testing of an approach to anti-islanding in utilityinterconnected photovoltaic systems,” Sandia National Laboratories, Albuquerque, NM, Tech. Rep. SAND 2000-1939, 2000. [Online]. Available: http://photovoltaics.sandia.gov/docs/PDF/0800steve.pdf [12] W. Wang, J. Kliber, G. Zhang, W. Xu, B. Howell, and T. Palladino, “A power line signaling based scheme for anti-islanding protection of distributed generatorspart ii: Field test results,” Power Delivery, IEEE Transactions on, vol. 22, no. 3, pp. 1767–1772, July 2007. [13] M. Ropp, M. Begovic, and A. Rohatgi, “Analysis and performance assessment of the active frequency drift method of islanding prevention,” IEEE Transaction on Energy Conversion, vol. 14, no. 3, pp. 810–816, 1999. [14] IEEEStandards, “Ieee standard conformance test procedures for equipment interconnecting distributed resources with electric power systems,” IEEE Std 1547.1-2005, vol. 1, pp. 25–29, 2005. [15] H. Haberlin and J. Graf, “Islanding of grid-connected pv inverters: Test circuits and some test results,” in Proc. 2nd World Conf. Exhibition on Photovoltaic Solar Energy Conversion, Vienna, Austria, 1998, pp. 2020–2023. [Online]. Available: http://labs.hti.bfh.ch/ fileadmin/user upload/lab1/pv/ilt982.pdf [16] H. Kobayashi, “Islanding prevention - evaluation test,” IEA PVPSTask V: Workshop about Grid Interconnection of Photovoltaic Power Systems, Zurich, Switzerland, Tech. Rep., 1997. 96 Bibliography [17] A. Woyte, R. Belmans, and J. Nijs, “Testing the islanding protection function of photovoltaic inverters,” IEEE Transaction on Energy Conversion, vol. 18, no. 1, pp. 157–162, 2003. [18] H. Jin. [Online]. Available: www.powersimtech.com [19] Y. Jung, J. Sol, G. Yu, and J. Choj, “Modelling and analysis of active islanding detection methods for photovoltaic power conditioning systems,” in Proc. Canadian Conference on Electrical and Computer Engineering, vol. 2, 2004, pp. 979–982 Vol.2. [20] M. Wrinch, J. Marti, and M. Nagpal, “Negative sequence impedance based islanding detection for distributed generation (nsiid),” in Electric Power Conference, 2008. EPEC 2008. IEEE Canada, Oct. 2008, pp. 1–6. [21] V. Blasko and V. Kaura, “A new mathematical model and control of a three-phase ac-dc voltage source converter,” IEEE Transactions on Power Electronics, vol. 12, no. 1, pp. 116–123, January 1997. [22] J.-C. Liao and S.-N. Yeh, “A novel instantaneous power control strategy and analytic model for integrated rectifier/inverter systems,” Power Electronics, IEEE Transactions on, vol. 15, no. 6, pp. 996–1006, Nov 2000. [23] D. Casadei, G. Serra, and K. Tani, “Implementation of a direct control algorithm for induction motors based on discrete space vector modulation,” Power Electronics, IEEE Transactions on, vol. 15, no. 4, pp. 769–777, Jul 2000. [24] F. Alberto, B. Batista, and I. Barbi, “Space vector modulation applied to three-phase three-switch two-level unidirectional pwm rectifier,” in Proc. Twenty-First Annual IEEE Applied Power Electronics Conference and Exposition APEC ’06, 2006, pp. 7 pp.–. [25] T. Basso and R. DeBlasio, “Ieee 1547 series of standards: interconnection issues,” vol. 19, no. 5, pp. 1159–1162, 2004. [26] M. Brenna, G. Lazaroiu, G. Superti-Furga, and E. Tironi, “Bidirectional front end converter for dg with disturbance insensitivity and islanding-detection capability,” vol. 23, no. 2, pp. 907–914, 2008. [27] B. Yu, Y. Jung, J. So, H. Hwang, and G. Yu, “A robust anti-islanding method for grid-connected photovoltaic inverter,” in Photovoltaic 97 Bibliography Energy Conversion, Conference Record of the 2006 IEEE 4th World Conference on, vol. 2, May 2006, pp. 2242–2245. [28] Z. Chen, F. Blaabjerg, and J. K. Pedersen, “A multi-functional power electronic converter in distributed generation power systems,” in Proc. IEEE 36th Power Electronics Specialists Conf. PESC ’05, 2005, pp. 1738–1744. [29] Z. Chen, Y. Hu, F. Blaabjerg, and J. K. Pedersen, “Stability improvement with a multi-functional power electronic converter in distributed power systems,” in Proc. IECON 2006 - 32nd Annual Conf. IEEE Industrial Electronics, 2006, pp. 5167–5172. [30] G.-H. Choe, H.-S. Kim, H.-G. Kim, Y.-H. Choi, and J.-C. Kim, “The characteristic analysis of grid frequency variation under islanding mode for utility interactive pv system with reactive power variation scheme for anti-islanding,” in Proc. 37th IEEE Power Electronics Specialists Conference PESC ’06, 2006, pp. 1–5. [31] J. Choi, Y. Jung, and G. Yu, “Novel afd method with pulsation of chopping fraction for islanding prevention of grid-connected photovoltaic inverter,” in Proc. European Conference on Power Electronics and Applications, 2005, pp. 10 pp.–P.10. [32] S.-K. Chung, “A phase tracking system for three phase utility interface inverters,” Power Electronics, IEEE Transactions on, vol. 15, no. 3, pp. 431–438, May 2000. [33] M. Ciobotaru, V. Agelidis, and R. Teodorescu, “Accurate and lessdisturbing active anti-islanding method based on pll for grid-connected pv inverters,” in Proc. IEEE Power Electronics Specialists Conference PESC 2008, 2008, pp. 4569–4576. [34] D. Dal and N. Mansouri, “Determining the optimal number of islands in power islands synthesis,” in Proc. IEEE Computer Society Annual Symposium on VLSI ISVLSI ’08, 2008, pp. 22–27. [35] F. Delfino, G. B. Denegri, M. Invernizzi, R. Procopio, and G. Ronda, “A p-q capability chart approach to characterize grid connected pv-units,” in Proc. CIGRE/IEEE PES Joint Symp. Integration of Wide-Scale Renewable Resources Into the Power Delivery System, 2009, pp. 1–8. 98 Bibliography [36] L. Ding, Z. Pan, W. Cong, and J. Pang, “An integrated automatic control system for distributed generation hierarchical islanding,” in Proc. International Conference on Power System Technology PowerCon 2006, 2006, pp. 1–6. [37] M. El-werfelli, J. Brooks, and R. Dunn, “Controlled island ng scheme for power systems,” in Proc. 43rd International Universities Power Engineering Conference UPEC 2008, 2008, pp. 1–6. [38] W. Feng, “A dsp based ac electronic load for unintentional islanding tests,” Electronic Thesis or Dissertation, University of British Columbia, November 2009. [Online]. Available: http: //hdl.handle.net/2429/11991 [39] N. Fukuoka, E. Nakashima, Y. Miyamoto, and H. Sugihara, “The simulated equipment of distribution line for the research on clustered pv systems,” in Conference Record of the 2006 IEEE 4th World Conference on Photovoltaic Energy Conversion, vol. 2, 2006, pp. 2411–2414. [40] A. Ghosh and G. Ledwich, “Load compensating dstatcom in weak ac systems,” Power Delivery, IEEE Transactions on, vol. 18, no. 4, pp. 1302–1309, Oct. 2003. [41] R. Hudson, T. Thorne, F. Mekanik, M. Behnke, S. Gonzalez, and J. Ginn, “Implementation and testing of anti-islanding algorithms for ieee 929-2000 compliance of single phase photovoltaic inverters,” in Conference Record of the Twenty-Ninth IEEE Photovoltaic Specialists Conference, 2002, pp. 1414– 1419. [Online]. Available: http://ieeexplore.ieee.org/stamp/stamp. jsp?arnumber=1190874&isnumber=26685 [42] C. Jeraputra and P. Enjeti, “Development of a robust anti-islanding algorithm for utility interconnection of distributed fuel cell powered generation,” vol. 19, no. 5, pp. 1163–1170, 2004. [43] N. P. Johansson, H.-P. Nee, and L. Angquist, “Estimation of grid parameters for the control of variable series reactance facts devices,” in Proc. IEEE Power Engineering Society General Meeting, 2006. [44] V. John, E. Benedict, and S. Danial, “An universal interconnection system to connect distributed generation to the grid,” in Proc. International Conference on Power Electronics, Drives and Energy Systems PEDES ’06, 2006, pp. 1–7. 99 Bibliography [45] B. Johnson, “Implementing ieee 1547 as an interconnection standard in pjm,” in Proc. IEEE PES Power Systems Conference and Exposition, 2004, pp. 1641–1644 vol.3. [46] V. John, Z. Ye, and A. Kolwalkar, “Investigation of anti-islanding protection of power converter based distributed generators using frequency domain analysis,” Power Electronics, IEEE Transactions on, vol. 19, no. 5, pp. 1177–1183, Sept. 2004. [47] H. Karimi, A. Yazdani, and R. Iravani, “Negative-sequence current injection for fast islanding detection of a distributed resource unit,” Power Electronics, IEEE Transactions on, vol. 23, no. 1, pp. 298–307, Jan. 2008. [48] M. Kazerani, “A high-performance controllable ac load,” in Proc. 34th Annual Conf. of IEEE Industrial Electronics IECON 2008, 2008, pp. 442–447. [49] A. Kitamura, M. Okamoto, K. Hotta, K. Takigawa, H. Kobayashi, and Y. Ariga, “Islanding prevention measures: demonstration testing at rokko test center for advanced energy systems,” in Conference Record of the Twenty Third IEEE Photovoltaic Specialists Conference, 1993, pp. 1063–1067. [50] D. Klapp and H. Vollkommer, “Application of an intelligent static switch to the point of common coupling to satisfy ieee 1547 compliance,” in Proc. IEEE Power Engineering Society General Meeting, 2007, pp. 1–4. [51] M.-J. Ko, I. Choy, J.-Y. Choi, K.-O. Lee, and G.-J. Yu, “Simulation of active frequency drift adding zero current method for islanding detection,” in Power Electronics Specialists Conference, 2006. PESC ’06. 37th IEEE, June 2006, pp. 1–5. [52] L. Lopes and H. Sun, “Performance assessment of active frequency drifting islanding detection methods,” Energy conversion, ieee transactions on, vol. 21, no. 1, pp. 171–180, March 2006. [53] K. Malmedal, B. Kroposki, and P. Sen, “Energy policy act of 2005,” vol. 13, no. 1, pp. 14–20, 2007. [54] K. Malmedal and P. Sen, “Comparison of some randomly selected utilities interconnection requirements and the compliance with the ieee std. 100 Bibliography 1547 - interconnection guidelines,” in Proc. IEEE Rural Electric Power Conference, 2008, pp. C3–C3–8. [55] R. Mastromauro, M. Liserre, and A. Dell’Aquila, “Frequency domain analysis of inductor saturation in current controlled grid converters,” in Proc. 33rd Annual Conference of the IEEE Industrial Electronics Society IECON 2007, 2007, pp. 1396–1401. [56] M. K. Mishra, A. Ghosh, A. Joshi, and H. M. Suryawanshi, “A novel method of load compensation under unbalanced and distorted voltages,” Power Delivery, IEEE Transactions on, vol. 22, no. 1, pp. 288– 295, Jan. 2007. [57] M. Molinas and J. Kondoh, “Power electronic loads as providers of reactive power ancillary service to the grid: Analytical and experimental study,” in Proc. 13th European Conf. Power Electronics and Applications EPE ’09, 2009, pp. 1–10. [58] S. S. Mortazavi, R. Kianinezhad, and A. Ghasemi, “Auto tuned robust active power filter for power quality improvement under fast load variation,” in Proc. Third Int. Conf. Electric Utility Deregulation and Restructuring and Power Technologies DRPT 2008, 2008, pp. 1646– 1651. [59] V. Rajput, S. Rovnyak, S. Koskie, and Y. Sheng, “A microcontrollerbased phasor measurement system with can bus communication,” in Proc. IEEE Power and Energy Society General Meeting - Conversion and Delivery of Electrical Energy in the 21st Century, 2008, pp. 1–5. [60] M. Reza, P. Schavemaker, J. Slootweg, W. Kling, and L. van der Sluis, “Impacts of distributed generation penetration levels on power systems transient stability,” in Proc. IEEE Power Engineering Society General Meeting, 2004, pp. 2150–2155 Vol.2. [61] M. Robitaille, K. Agbossou, and M. Doumbia, “Modeling of an islanding protection method for a hybrid renewable distributed generator,” in Proc. Canadian Conference on Electrical and Computer Engineering, 2005, pp. 1477–1481. [62] M. Ropp, M. Begovic, A. Rohatgi, G. Kern, S. Bonn, R.H., and S. Gonzalez, “Determining the relative effectiveness of islanding detection methods using phase criteria and nondetection zones,” IEEE Transaction on Energy Conversion, vol. 15, no. 3, pp. 290–296, 2000. 101 Bibliography [63] M. Ropp, D. McMahon, and K. Fennell, “Estimating the realistic ranges of load parameters for anti-islanding testing,” in Conference Record of the Thirty-first IEEE Photovoltaic Specialists Conference, 2005, pp. 1765–1768. [Online]. Available: http://ieeexplore.ieee.org/search/srchabstract.jsp?arnumber= 1488492&isnumber=31426&punumber=9889&k2dockey=1488492@ ieeecnfs&query=((ropp+m)%3Cin%3Emetadata)&pos=2&access=no [64] M. Ropp, J. Ginn, J. Stevens, W. Bower, and S. Gonzalez, “Simulation and experimental study of the impedance detection anti-islanding method in the single-inverter case,” in Conference Record of the 2006 IEEE 4th World Conference on Photovoltaic Energy Conversion, vol. 2, 2006, pp. 2379–2382. [65] S. Rovnyak, S. Koskie, Y. Sheng, V. Rajput, and X. Hu, “Laboratory testing of a distributed generator unintentional islanding detection scheme,” in Proc. IEEE Power and Energy Society General Meeting Conversion and Delivery of Electrical Energy in the 21st Century, 2008, pp. 1–6. [66] R. Saint and N. Friedman, “The application guide for distributed generation interconnection-the nreca guide to ieee 1547,” in Proc. IEEE Rural Electric Power Conference, 2002, pp. D2–D2,5. [67] C. Schauder and H. Mehta, “Vector analysis and control of advanced static var compensators,” Generation, Transmission and Distribution, IEE Proceedings C, vol. 140, no. 4, pp. 299–306, Jul 1993. [68] J. Slootweg and W. Kling, “Impacts of distributed generation on power system transient stability,” in Proc. IEEE Power Engineering Society Summer Meeting, vol. 2, 2002, pp. 862–867 vol.2. [69] H.-S. Song and K. Nam, “Dual current control scheme for pwm converter under unbalanced input voltage conditions,” Industrial Electronics, IEEE Transactions on, vol. 46, no. 5, pp. 953–959, Oct 1999. [70] N. R. Ullah, “Small scale integration of variable speed wind turbines into the local grid and its voltage stability aspects,” in Proc. Int Future Power Systems Conf, 2005. 102 [71] W. Xiaoxi, L. Lin, L. Fen, and Z. Yunping, “Design of single-phase ac power electronic load,” in Proc. Int Electrical and Control Engineering (ICECE) Conf, 2010, pp. 5843–5846. [72] A. Yazdani and R. Iravani, “A unified dynamic model and control for the voltage-sourced converter under unbalanced grid conditions,” Power Delivery, IEEE Transactions on, vol. 21, no. 3, pp. 1620–1629, July 2006. [73] B. Yu, Y. Jung, J. So, H. Hwang, and G. Yu, “A robust anti-islanding method for grid-connected photovoltaic inverter,” in Conference Record of the 2006 IEEE 4th World Conference on Photovoltaic Energy Conversion, vol. 2, 2006, pp. 2242–2245. [74] H. Zeineldin and J. Kirtley, “A simple technique for islanding detection with negligible nondetection zone,” Power Delivery, IEEE Transactions on, vol. 24, no. 2, pp. 779–786, April 2009. [75] Z. Zhi, X. Yun-xiang, H. Wei-ping, L. Jiang-yuan, and C. Lin, “A new svpwm method for single-phase three-level npc inverter and the control method of neutral point voltage balance,” pp. 1 –4, 2009. 103 Appendix A Space Vector PWM A.1 Three-Phase PWM Inverter Model Q1 Q3 Q5 PVDG & VA Vdc VB VC DC/DC Converter Q2 Q4 Q6 Figure A.1: Three-phase two-leve inverter Space vector pulse width modulation (SVPWM) will be used to control a three phase two level inverter as in figure A.1. SVPWM principle is based on vector representation of PWM inverter output voltage generated from all possible switching configurations. SVPWM outperforms the conventional sinusoidal PWM as it improves on output gain, less switching loss, and less output ripple. The application of SVPWM to control the inverter requires calculating switching state ON-time duration. Two solutions are presented. In section A.2, the derivation solves ON-time durations using sampled reference space vector phase-angle. In section A.3, the derivation solves ON-time durations using sampled reference phase voltage amplitudes. In section A.4, 104 A.2. Sampled Space Vector Phase Angle U⃗3 S2 U⃗2 S3 S1 t2 v⃗s∗ α U⃗4 t1 U⃗1 S6 S4 U⃗5 S5 U⃗6 Figure A.2: Sectors of SVPWM Simulation results for both solutions are presented and compared. A.2 Sampled Space Vector Phase Angle In figure A.2, an arbitrary rotating reference space vector ⃗vs∗ is placed in sector S1 ; where (∗ ) above ⃗vs denotes a reference quantity. To represent ⃗vs∗ using the adjacent space vectors representing switch positions, ON-time ⃗ 1 and U ⃗ 2 respectively must be solved. duration t1 and t2 for space vectors U While the space vector phase angle alpha in figure A.2 is defined as the angle starting counterclockwise from the space vector in the sector to the reference voltage space vector ⃗vs∗ , a solution for t1 and t2 is derived in terms of alpha. A.2.1 Solution for t1 and t2 in Sector S1 Space vector ⃗vs∗ in sector S1 of figure A.2, can be described as |⃗vs∗ | (cos (α) + i sin (α)) = (π ) ( π )] t1 ⃗ t2 ⃗ [ |U1 | + |U + i sin 2 | cos ts ts 3 3 (A.1) ⃗ 1 and t2 where ts is the sampling period, t1 is the time duration for vector U ⃗ 2 . To solve for t2 , we use the imaginary is the ON-time duration for vector U 105 A.2. Sampled Space Vector Phase Angle part of equation A.1. (π ) ( π )) t1 ⃗ t2 ⃗ ( |U1 | + |U | cos + ı sin } 2 ts ts 3 3 (π ) t2 ⃗ |⃗vs∗ | sin (α) = |U 2 | sin ts 3 ℑ{|⃗vs∗ | (cos (α) + ı sin (α)) = solving for t2 yields t2 = ts |⃗vs∗ | sin (α) ( ) ⃗ 2 | sin π3 |U (A.2) substituting A.2 into the real part of equation A.1 yields (π ) ( π )) t2 ⃗ ( t1 ⃗ | + | U | ℜ{|⃗vs∗ |(cos (α) + i sin (α)) = |U cos + i sin } 1 2 ts ts 3 3 (π ) t1 ⃗ t2 ⃗ |⃗vs∗ | cos (α) = |U |U2 | cos 1| + ts ts 3 .. . .. . ( ) ⃗ 1 | = ts |⃗vs∗ | cos (α) − ts |⃗vs∗ | sin ((α)) cos π t1 |U 3 sin π3 Simplifying and solving for t1 yields ) (π) (α) 3 (sin ) cos (α) − sin π3 ( ) ) ( ( ) |⃗vs∗ | cos (α) sin π3 − cos π3 sin (α) ( ) t1 = ts ⃗ 1| sin π3 |U ( ) |⃗vs∗ | sin π3 − α ( ) t1 = ts (A.3) ⃗ 1 | sin π3 |U |⃗v ∗ | t1 = ts s ⃗ 1| |U ( cos ⃗ 0 and U ⃗ 7 is and the ON-time duration for the zero-state space vectors U t0 = t7 = A.2.2 ts − t1 − t2 2 (A.4) Solving for t2 and t3 in Sector S2 ⃗ 1 and U ⃗ 2 in Knowing the form of the ON-time duration for space vector U ⃗ ⃗ sector S1 , ON-time durations t2 and t3 for space vectors U2 and U3 respectively will be derived next. 106 A.2. Sampled Space Vector Phase Angle U⃗3 U⃗2 v⃗s∗ t3 t2 α2 α U⃗1 Figure A.3: Sector 2 For a space vector ⃗vs∗ in sector S2 as in figure A.3, it can be described as [ ( ) ( )] (π ) ( π )] t t2 ⃗ [ 2π 2π 3 ⃗ ∗ |⃗vs | (cos (α) + i sin (α)) = |U2 | cos + i sin + |U3 | cos + i sin ts 3 3 ts 3 3 (A.5) taking the real part of equation A.5 and solving for t3 yields ( ) (π ) t 2π t2 ⃗ 3 ⃗ ∗ + |U3 | cos |⃗vs | cos (α) = |U2 | cos ts 3 ts 3 ) ( ⃗ 2 | cos π ts |⃗vs∗ | cos (α) − t2 |U 3 t3 = (A.6) ( 2π ) ⃗ 3 | cos |U 3 while the imaginary part of equation A.5 is ( ) (π ) t 2π t2 ⃗ 3 ⃗ ∗ + |U3 | sin |⃗vs | sin (α) = |U2 | sin ts 3 ts 3 solving for t2 yields ⃗ 3 | sin ts |⃗vs∗ | sin (α) − t3 |U t2 = (π) ⃗ 2 | sin |U ( 2π ) 3 3 = ts |⃗vs∗ | sin (α) ( ) ⃗ 2 | sin π |U 3 ( ) ⃗ 3 | sin 2π t3 |U − ( 3) ⃗ 2 | sin π |U 3 substituting t3 from equation A.6 and solving for t2 will result in .. . |⃗v ∗ | t2 = ts s ⃗ 2| |U ( ( ) (π)) − cos (α) sin 1 sin (α) cos 2π 3 (3 ) ( ) 2 sin π3 cos 2π 3 107 A.2. Sampled Space Vector Phase Angle ( ) ( ) ( ) ( ) since sin π3 = sin 2π and cos 2π = − cos π3 = − 12 , t2 can be further 3 3 simplified to ( ( ) ( )) |⃗vs∗ | sin (α) cos π3 + cos (α) sin π3 ( ) = ts ⃗ 2| sin π3 |U ) ( |⃗vs∗ | sin α + π3 ( ) t2 = ts ⃗ 2 | sin π3 |U (A.7) ⃗ 2 ) to the Since the definition of α is from the start of the sector (from U π ∗ reference space vector (to ⃗vs ), α2 is defined as equal to α − 3 , therefore ( ) |⃗vs∗ | sin α2 + 2π 3 ( ) t2 = ts ⃗ 2| sin π3 |U .. . ) ( |⃗vs∗ | sin π3 − α2 ( ) = ts ⃗ 2 | sin π3 |U which is equivalent in form to t1 of sector S1 as in equation A.3. To solve for t3 , substitute t2 from equation A.7 in equation A.6 and solve ( ) ⃗ 2 | cos π |⃗vs∗ | cos (α) |U 3 ) ( 2π ) − t2 ( 2π t3 = ts ⃗ ⃗ cos cos |U3 | |U3 | 3 ( 3 π) ∗ ∗ ⃗ 2| |⃗v | cos (α) |⃗v | sin α + 3 |U (π) ( 2π ) + ts s = ts s ⃗ 3 | cos 3 ⃗ 2 | sin 3 ⃗ 3| |U |U |U .. . .. . ) ( |⃗vs∗ | sin α − π3 ( ) = ts ⃗ 3 | sin π3 |U To write t3 in terms of α2 results in t3 = ts |⃗vs∗ | sin (α2 ) ( ) ⃗ 3 | sin π3 |U (A.8) which is also equivalent in form to t2 of sector S1 as in equation A.2 108 A.3. Sampled Phase Voltage Amplitudes Continuing to solve for states ON-time durations for the all sectors will result in the same form of solution but with the respective sector’s phase angle. Therefore, defining t1 to be the ON-time duration for the start of the sector and t2 for the ON-time duration for the end of the sector, we can solve for any sector’s ON-time duration knowing the sector number and phase angle α. For sectors S1 , S3 and S5 ( ) |⃗vs∗ | sin π3 − α ( ) t1 = ts ⃗ 1 | sin π3 |U t2 = ts |⃗vs∗ | sin (α) ( ) ⃗ 2 | sin π3 |U (A.9) For sectors S2 , S4 and S6 |⃗vs∗ | sin (α) ( ) ⃗ 2 | sin π3 |U ( ) |⃗vs∗ | sin π3 − α ( ) t2 = ts ⃗ 1 | sin π3 |U t1 = ts (A.10) refer to table A.1 for phase angle α range in each sector. A.3 Sampled Phase Voltage Amplitudes In this section a stationary (α, β) transformation of the phase voltages of figure A.4 and space vectors of figure A.2 is carried out. Since the reference space vector ⃗vs∗ of figure A.5 will be described in terms of α and β, an inverse transformation will discribe the reference space vector in terms of phase voltage amplitudes. We define Vα and Vβ as Vα = |⃗vs∗ | cos (α) Vβ = |⃗vs∗ | sin (α) (A.11) then in terms of the phase voltages, Vα = VA + VB cos 2π −2π + VC cos 3 3 Vβ = VB sin 2π −2π + VC sin (A.12) 3 3 109 A.3. Sampled Phase Voltage Amplitudes Table A.1: ON-time durations for space vector PWM Sector (ABC) 1 (100) 2 (110) 3 (010) 4 (011) 5 (001) 6 (101) t1 ∗ ts |⃗v⃗s | sin ( π3 −α) sin ( π3 ) |U1 | ∗ sin α2 ts |⃗v⃗s | sin |U2 | ( π3 ) ∗ ts |⃗v⃗s | sin ( π3 −α3 ) sin ( π3 ) | U1 | ∗ sin α4 ts |⃗v⃗s | sin |U2 | ( π3 ) ∗ ts |⃗v⃗s | sin ( π3 −α5 ) | U1 | sin ( ∗ π 3 ) sin α6 ts |⃗v⃗s | sin |U2 | ( π3 ) t2 αi ts |⃗v⃗s | sinsin απ |U2 | (3) α=α ∗ ∗ ts |⃗v⃗s | sin ( π3 −α2 ) sin ( π3 ) |U3 | ∗ sin α3 ts |⃗v⃗s | sin |U2 | ( π3 ) ∗ ts |⃗v⃗s | sin (α4 − π3 ) sin ( π3 ) |U3 | ∗ sin α5 ts |⃗v⃗s | sin |U2 | ( π3 ) ∗ ts |⃗v⃗s | |U3 | sin ( π3 −α6 ) sin ( π3 ) range 0≤α< α2 = α − π 3 π 3 ≤α< α3 = α − 2π 3 2π 3 π 3 2π 3 ≤α<π α4 = α + π −π ≤ α < − 2π 3 α5 = α + 2π 3 π − 2π 3 ≤ α < −3 α6 = α + π 3 − π3 ≤ α < 0 110 A.3. Sampled Phase Voltage Amplitudes + Q1 VDC Q3 VA Q2 Q5 VB Q4 VC Q6 Figure A.4: Three-phase VSI but for a balanced load three-phase system 0 = VA + VB + VC VA = −(VB + VC ) (A.13) simplifying equation A.12 using equation A.13 yields Vα = VA + cos 2π (VB + VC ) 3 1 = VA + VA 2 3 Vα = VA 2 Vβ = sin 2π (VB − VC ) 3 √ 3 Vβ = (VB − VC ) 2 (A.14) The maximum amplitude for ⃗vs∗ is limited to the inner tangent circle of figure A.5 and can be easily calculated at point A where α = 30o . ( π ) √3 Vα |M AX = U1 cos = Vdc (A.15) 6 2 where the magnitude of the space vectors is equal to Vdc . From A.14 and A.15 we can find the maximum phase voltage VA √ Vdc 2 3 Vdc = √ (A.16) VA |M AX = 3 2 3 111 A.3. Sampled Phase Voltage Amplitudes ⃗β V ⃗2 U ⃗ 2β U b ⃗vsβ ⃗vs∗ t2 α A t1 ⃗ 2α ⃗vsα U ⃗1 U ⃗α V Figure A.5: (V⃗α ,V⃗β ) transformation in sector S1 Referring to equation A.3 ( ) |⃗vs∗ | sin π3 − α ( ) t1 = ts ⃗ 1 | sin π3 |U (π ) 2 |⃗v ∗ | = ts √ s sin −α 3 3 Vdc (π ) ( π )) 2 ts ( ∗ =√ |⃗vs | cos (α) sin − |⃗vs∗ | sin (α) cos 3 3 3 Vdc (π ) ( π )) 2 ts ( Vα sin =√ − Vβ cos 3 3 3 Vdc substituting equation A.14 for Vα and Vβ (√ ) √ 2 ts 33 1 3 =√ VA − (VB − VC ) 2 2 2 2 3 Vdc ( ) ts 3 1 1 = VA − VB + VC Vdc 2 2 2 ( ) ts 1 1 1 = VA − VB + VA + VC Vdc 2 2 2 (A.17) (A.18) from equation A.13 we can replace 1/2VA + 1/2VB with −1/2VB = ts (VA − VB ) Vdc (A.19) 112 A.3. Sampled Phase Voltage Amplitudes Therefore, in terms of phase voltage amplitudes t1 can be described as t1 = ts (TA − TB ) (A.20) where TA = VA Vdc TB = VB Vdc TC = VC Vdc (A.21) are time ratios of ts . t2 = ts |⃗vs∗ | sin (α) ( ) ⃗ 1 | sin π3 |U 2 |⃗v ∗ | = ts √ s sin (α) 3 Vdc 2 ts ∗ =√ |⃗vs | sin (α) 3 Vdc 2 ts =√ Vβ 3 Vdc substituting equation A.14 for Vβ √ 2 ts 3 =√ (VB − VC ) 3 Vdc 2 ts = (VB − VC ) Vdc t2 = ts (TB − TC ) (A.22) (A.23) where TB and TC as in equation A.21 For sector 2, (π ) 2 |⃗v ∗ | t1 = ts √ s sin − α2 3 3 Vdc ( (π ) ( π )) ∗ 2 |⃗v | = ts √ s cos α2 sin − sin α2 cos 3 3 3 Vdc (π ) ( π )) 2 ts ( ∗ =√ |⃗vs | cos α2 sin − |⃗vs∗ | sin α2 cos 3 3 3 Vdc (A.24) 113 A.3. Sampled Phase Voltage Amplitudes Vβ Vα2 U⃗3 v⃗s∗ U⃗2 Vβ2 α2 t1 t2 30o α Vα Figure A.6: Sector 2 α,β transformation |⃗vs∗ | cos α2 and |⃗vs∗ | sin α2 are the Vα2 and Vβ2 components as in figure A.3. To get t1 in terms of α and β we find the contribution of Vα and Vβ to sector 2 Vα2 -Vβ2 coordinates. √ (π ) 1 (π ) 3 ∗ + Vβ cos = Vα + Vβ Vα2 = |⃗vs | cos α2 = Vα cos 3 6 2 √ 2 (π ) (π ) 1 3 Vβ2 = |⃗vs∗ | sin α2 = −Vα cos Vα + Vβ + Vβ cos =− 6 3 2 2 from the above, we can solve t1 and t2 of sector 2 as follows [( ) ) ] (√ √ (π ) (π ) 2 ts 1 3 3 1 t1 = √ Vα + Vβ sin Vα − Vβ cos + 2 2 3 2 2 3 3 Vdc [√ ( ) )] (√ √ 1 3 1 3 3 1 2 ts Vα + Vβ + Vα − Vβ =√ 2 2 2 2 2 3 Vdc 2 [√ ] √ 2 ts 3 1 3 3 =√ Vα + Vβ + Vα − Vβ V 4 4 4 4 3 dc [√ ] 2 ts 3 1 =√ Vα + Vβ V 2 2 3 dc substituting equation A.14 for Vα and Vβ [√ ] √ 33 2 ts 1 3 =√ VA + (VB − VC ) 2 2 3 Vdc 2 2 [ ] 1 1 ts 3 VA + VB − VC = Vdc 2 2 2 114 A.4. Space Vector PWM ON-Time Durations for Three-Phase Inverter the same elimination as in equation A.18 we get the final form for t1 t1 = ts (TA − TC ) (A.25) and t2 is 2 |⃗v ∗ | t1 = ts √ s sin α2 3 Vdc [ √ ] 3 1 2 ts − Vα + Vβ =√ 2 2 3 Vdc substituting equation A.14 for Vα and Vβ = = = = [ √ ] √ 1 3 33 2 ts √ − VA + (VB − VC ) 2 2 2 2 3 Vdc [ ] 3 1 ts − VA + (VB − VC ) Vdc 2 2 [ ] ts 1 1 1 −VA + VB − VC − VA Vdc 2 2 2 ts [VB − VA ] Vdc and t2 for sector 2 will be t2 = ts (TB − TA ) (A.26) Table A.2 describes the ON-time durations for the SVPWM using sampled phase voltage amplitudes A.4 Space Vector PWM ON-Time Durations for Three-Phase Inverter To synthesize the reference space vector voltage ⃗vs∗ from active and zero switching states of the three phase inverter in figure A.4, ON-time durations are expressed as tk = |Uk | ts Vdc k = 1, 2, ..6 (A.27) 115 A.4. Space Vector PWM ON-Time Durations for Three-Phase Inverter Table A.2: Fundamental space vectors ONtime duration described using sampled phase voltage amplitudes Sector 1 2 3 4 5 6 (Q1, Q2, Q3) 1, 0, 0 1, 1, 0 0, 1, 0 0, 1, 1 0, 0, 1 1, 0, 1 ts1 * TA − TB TA − TC TB − TC TB − TA TC − TA TC − TB ts2 * TB − TC TB − TA TC − TA TC − TB TA − TB TA − TC * t =t ·t , t =t ·t s s 1 s1 2 s2 while the rest of the smapling period will be devided evenly between the zero states tz = t000 = t111 = ts − t1 − t2 2 (A.28) + T1 Vdc VA T2 T3 VB T4 T5 VC T6 Figure A.7: Three-phase 116 Appendix B Hardware Implementation After arriving at the results of the simulation and design of the island stabilizing element. Hardware is built based on the theoretical design and in this chapter the approach to hardware design is discussed in detail and preliminary experimental results are shown and analysed. 2 Voltage, Current and Temperature Signals Isolated IGBT Driver Sensor Circuits Power Supply DSP Control Board Figure B.1: ISE hardware layout[38] From the block diagram above, the DC supply will be chosen according to the power demand needed. The H-Bridge inverter block will be discussed in details along with the IGBT driver circuit and DSP controller. The sensor circuit will provide the information needed by the DSP program to calculate the current of the inverter and also the sensors will provide protection infor117 Appendix B. Hardware Implementation mation like temperature to prevent damage to the circuit. The current filter will also be discussed in detail in a section of its own. Finally, the IGBT gating signals will need DC (± 15V) power supplies which are shown as an external auxiliary block to the main system, as well as the sensors would require some independent DC power supplies (+15V, +5V). B.0.1 Inverter Using IGBT Model MUBW 20-06 A7 The IGBT model is employed to provide the switching signals for the HBridge single phase inverter used as an island stabilizing element. 21 D11 D13 22 D7 D15 7 1 2 3 D12 D14 D16 T7 14 23 T1 16 15 T2 11 10 D1 6 T3 18 17 T4 D2 12 D3 5 D4 T5 D5 20 19 4 T6 D6 13 24 8 NTC 9 Figure B.2: IGBT module [38] The IGBT model has a three phase diode rectifier and a three phase IGBT inverter H-bridge circuit. It also include an NTC temperature sensor. The ISE hardware design would need only the inverter side of the module in addition to the NTC temperature sensor. Provided that the DC power supply is readily available. Since the ISE would be used for equipment meant for connection to the grid at the distribution level, the voltage requirement is 120 Vrms at frequency of 60 Hz. The current rating for the ISE would need to be equal or higher than 8.33 Arms as it corresponds to 1kW of power or higher. The IGBT module is rated up to 600 V and 35 A which meets the minimum 118 Appendix B. Hardware Implementation requirement. To protect the IGBT module from damage, the temperature sensor is used which is an NTC thermistor. The sensors’s resistance decrease with increasing temperature. The sensor’s main function will be one of protection of the IGBT block. B.0.2 Driver Circuit A compact IGBT driver module (6SD106EI) is used to provide the needed gating voltage signals for the IGBTs to switch properly. The driver module is capable of providing a high gate current of ± 6A. The driver module is capable of producing six independent signals for all six IGBTs or in a three-pair signal mode. Also, it can function as a full-bridge or half-bridge mode. The module also provides security from short-circuit and over-current built-in. Figure B.3 shows the driver circuit block diagram for one pair of IGBT switches. The driver has two output channels one for every IGBT. For the three-pair signals, there is only one PWM oscillator needed and all other components are duplicates for the other two IGBT legs. Each channel is optically isolated between power and control circuit to prevent any damage to the control circuit in case of failure in the power circuit. The internal IGD block in Figure B.3 is responsible for the over-current protection and shortcircuit prevention of the power transistor. Also, it had a feed-monitoring circuit and a status acknowledgement circuit. An independent ± 15 VDC power supply is available to provide the necessary IGBT driver voltage. B.0.3 Mode Selection Since the ISE is based on a single phase inverter design, the driver module will be operated in half-bridge mode with two pairs of the complimentary driver signals are needed to operate the four IGBTs. It is noted here that the driver module will output three pairs of complimentary IGBT drive signals. For half-bridge operation, pin MOD is connected to the GND reference. Also, input pins of RC1 and RC2 is connected to the RC network. The size of the RC network will be detailed through the dead time calculation. B.0.4 Dead Time To avoid short-circuit across two IGBT switched in one leg of the H-Bridge, a dead time, defined as the time where both IGBT switches are turned off in one leg, is included for protection. This dead time is necessary because 119 Appendix B. Hardware Implementation real IGBT switches can not turn on and turn off instantaneously. The dead time is set up either by using the half-bridge mode of the IGBT module or by including it in the program of the DSP controller. In the experimental set up, the half-bridge mode requires an RC circuit is required to set the dead time to 2.1us. The RC values for the desired dead time according to ?? are chosen to be 22kΩ and 150pF. B.0.5 Voltage Measuring Circuit The circuit for measuring the voltage at PCC consists of three stages. The voltage sensor, then a voltage level scaling, and finally a voltage buffer stage. The voltage sensor A voltage transducer was used to measure the voltage at PCC. The module, LV 25-P, can measure up to 500V which is proper for measuring 120 Vrms . The sensor outputs a current on the secondary side with turns ratio of 2.5:1 of the primary current. The sensor requires a DC voltage source of ± 15 V. The sensor circuit is shown in Figure B.4. The current output of the voltage sensor is proportional to the voltage being measured. A series resistor R1 = 6kΩ is used in series with the voltage sensor which will correspond to 10 mA primary current when the input voltage is set to 60V in the test experiment. In the islanding case, the input voltage is 120 Vrms which will require an R1 = 17 kΩ. Voltage scaling The EZDSPlf2407A DSP board used in the experiment has an ADC module which accepts input voltage range between 0 and 3.3V. Hence, the output voltage from the sensor needs to be scaled to a proper input voltage for the ADC. An operational amplifier is used in two stages to scale the voltage as in Figure B.5 120 Appendix B. Hardware Implementation Rth IGD VDD Rg Viso1 LDI Rth GND IGD Rg Viso2 VDC Viso1 PWM oscillator Viso2 GND Interface on Electronic Level Electrical Isolation Driver on Power Level Power Semiconductor (external) SCALE Driver Module Figure B.3: Block diagram of IGBT Module [38] 121 Appendix B. Hardware Implementation Figure B.4: The circuit of the voltage sensor [38] 122 -15i VT LV 25-P 3 2 1 R1 240 30K C3 30K 150pF +15i C1 1 11 C2 R4 12K 9 6 5 7 2 LM324AM U4B -15I 100nF -15i 1 3 NC 1 LM324AM U4A 2 U5 LM4040A30IDBZR-3.0 +15i +15I 4 2 3 3.3K 100nF 4 R6 4 5.1K R5 123 Figure B.5: Voltage scaling circuit [38] 10 3 11 -15i R3 R2 8 LM324AM U4C -15i VM1 Appendix B. Hardware Implementation +15i 11 CN7 Appendix B. Hardware Implementation The current signal from the voltage sensor converted to a voltage signal using R1 and with a rang of ± 6V. The input stage OpAmp, U4A, will produce and output voltage described as ( ) R3 R3 − VIN · (B.1) Vout1 = Vref 1 + R2 R2 In Equation B.1 VREF is defined as U5 = 3V, which is the voltage across the precision voltage reference LM4040A. To obtain the scaled output voltage Vout , the second OpAmp scales the input voltage Vout1 as Vout = − R5 · Vout1 R4 (B.2) To illustrate and summarize the scaling done in this circuit, Figure B.6 shows the stages of scaling signals from the primary current to DSP analog voltage input of the ADC. Current Sensor Offset IN Coefficient out1 out Figure B.6: Voltage scaling from sensor output current to DSP input ADC [38] Voltage buffer The ADC built in the DSP board is unbuffered multiplexed which contains a sample-and-hold circuit and an ADC comparator. This ADC does not employ internal buffer for input offset or gain error. In Figure B.7 shows the basic circuit of the ADC in the DSP board. One disadvantage of using such ADC is crosstalk. When the sample capacitor in the ADC is directly charged by the external signal, the charge is left on the capacitor in the 124 Appendix B. Hardware Implementation current sample might affect the accuracy of the next sample if inadequate sampling time is used. There is a minimum settling time for the ADC, which if not used, the conversion would contain sampling errors. This is referred to as crosstalk. sample MUX Figure B.7: Voltage buffer circuit [38] To eliminate such errors, design of the input impedance to the ADC should be much larger than the source impedance of the input signal. U4C in Figure B.5 is the voltage buffer added to reduce the output impedance of the scaling circuit. Current measuring circuit An LEM LT 100-S current transducer with a current range of ±100 A. The Transducer outputs a current with a ratio of 1:1000 of the primary current. In order to obtain accurate measurements, the primary current measured should be as close to 100A as possible. This is achieved with multiple turns of current being measured. Figure B.8 Shows the current measuring circuit. The parameters in the circuit are substituted into Equation B.1 and Equation B.2 will produce the scaled current graph shown in Figure B.9. 125 -15i CT LT 100-S 3 2 1 R1 62 C3 30K +15i 150pF C1 R6 2 1 +15i +15I 11 C2 R4 12K 9 6 5 7 2 LM324AM U1B -15I 100nF -15i 1 3 NC U5 LM4040A30IDBZR-3.0 1 LM324AM U1A 2 3 3.3K 100nF 4 4 5.1K R5 4 32K 126 Figure B.8: Current sensor circuit [38] 10 3 11 -15i R3 R2 8 LM324AM U1C -15i CM1 Appendix B. Hardware Implementation +15i 11 CN1 Appendix B. Hardware Implementation Current Sensor Offset IN Coefficient out1 out Figure B.9: Measured current scaling [38] Temperature circuit The NTC thermistor is integrated within the IGBT board for protection from over heating and damage to the switches. The data sheet indicate that the module has a range of -40 C o to 125 C o with a resistance of 300 Ω at the maximum temperature. The temperature is measured using a circuit that measures the thermistor resistance which varies with the temperature. In practical application, the recovery from over heating requires a period of time which can be avoided using a temperature hysteresis comparator circuit, Figure B.10 is used to ensure proper operation within allowed temperature range. When the temperature is below the maximum 125 C o , R16 > 300Ω, and the output voltage signal, U3A, is zero. Once the temperature reaches the maximum allowed, U3A will switch to a high signal of 5 V. Implementing the hysteresis temperature loop, R12 , the lower voltage limit can be calculated as VinM IN = VREF − R16 ∥ R15 · VH R16 ∥ R15 + R12 (B.3) Substituting the circuit parameters into the above equation to obtain VinM IN will yield a voltage of 2.35V. This means that Vout will only be zero when Vin is less than 2.35V rather than 2.5V. Figure B.11 shows the hysteresis temperature loop in which Vin is correlated to a resistance value from 300 Ω to 338 Ω. 127 Appendix B. Hardware Implementation +5 5K R12 3K R13 R16 12 1.5K R14 Vin 5 U3A 2 VREF Vout 4 3 LM339AM C31 2.5v D4 300 R15 +5 100nF Figure B.10: Temperature sensor circuit [38] DSP Control A DSP control is used to control the ISE based on TI’s TMS320LF2407A digital processor. The eZDSPLF2407A control board includes a JTAG connector for interface and debugging programs. The CPU speed is 40M Hz, 64kB RAM, 32kB ROM or Flash EEPROM, 64kB program, 64kB of data and finally 64kB of I/O space addressing. The board includes 16 multiplexed analog input 10-bit ADC with built-in S/H circuit. The conversion rate is 375ns. It supports 4 trigger sources for start-of-conversion sequence and auto-sequencing functions. For motor control applications, irrelevant here, the board includes an event manager (EV) module for a broad range of features that are useful in motion control applications. The board has 40 multiplexed I/O general purpose pins. For system reset, a watchdog timer is included for monitoring of software or hardware operation and implement system reset. A CAN, controller area network, is included on the board and a serial communication interface (SCI). It is noted here that the EV manager has a variety of features that are worth including in the description of the board. A Two general purpose timers, three general-purpose up and up/down timers are available. Each of these timers are 16-bit compare unit capable of generating an independent PWM output. The EV manager also includes a PWM circuit that can 128 Appendix B. Hardware Implementation ℃ ℃ ℃ Figure B.11: Temperature hysteresis loop [38] produce an SVPWM signals, dead-band generation and output logic. A three capture units are available via the EV manager as well as a QEP, a quadrature encoder pulse. Power Supply A commercial switching power supply, VOF-65-15, are used to supply the PCB with required power. Also, two commercial DC-DC converters were used to supply the required DC power needed by the circuit. A 5V DC-DC converter, CC10-1205SF-E, supplies the eZDSP board, temperature sensor and protection circuits. A ±15V DC power supply, CC10-1212DF-E, provides power for the OpAmp circuits and another power supply, VOF-65-15, feeds the driver circuit. Input filters and output smoothing capacitors are incorporated in the power supply circuits as recommended by the data sheet and application notes. The power supply circuit is shown in Appendix C. 129 Appendix C Circuit and PCB 130 C.1 Schematics of Control Board 1 2 3 4 CN1 +15 CC1 L1 1 3.5uH 10uF 10uF C6 10uF C7 10uF C8 10uF C9 10uF 10uF C3 C10 10uF C4 10uF C5 +Vin 10uF C11 C12 +Vout Trim 2 6 C2 C1 10uF 0.1uF C14 10uF C13 0.1uF 16V D1 RC Com 3 +15i 7 -Vin -Vout 5 4 16V D2 -15i CC10-1212DF-E +15 CC2 L2 1 3.5uH 10uF 10uF C20 10uF C21 10uF C22 10uF C23 C24 10uF 10uF C17 10uF C18 10uF C19 +Vin C25 C26 +Vout 7 +5 5V 10uF Trim 2 RC -Vout 3 6 -Vin NC 5 4 CC10-1205SF-E Figure C.1: PowerSupply AD.SCHDOC [38] C16 C15 10uF 0.1uF D3 C.1. Schematics of Control Board +15 131 C.1. Schematics of Control Board -15i 3 2 1 R16 Res1 240 R17 R18 Res1 30K -15i C32 30K +15i 150pF C33 R19 100nF 3.3K 2 1 1 LM324AM U4A 11 R21 12K 5 C34 100nF 2 9 6 7 2 10 LM324AM U4B 11 3 +15i +15I 4 4 5.1K R20 4 +15i 8 3 VM1 LM324AM U3C 11 CN7 -15i -15I U5 -15i LM4040A30IDBZR-3.0 -15i R22 Res1 240 CN8 +15i R23 R24 C35 30K 30K 150pF R25 8 3 R26 12K 13 13 12 -15I 14 4 12 LM324AM U4D 11 11 LM324AM U4C 14 4 VM2 LM324AM U3D 11 9 10 +15i +15I 4 4 3.3K 4 +15I -15i -15I -15i 3 2 1 R27 Res1 240 CN9 +15i R28 R29 C36 30K 150pF +15i +15i 4 +15I 1 1 LM324AM U6A C38 R31 12K 9 6 5 7 2 LM324AM U6B 10 8 3 VM3 LM324AM U6C 11 2 3 R30 3.3K 100nF 4 4 C37 11 30K 11 1 3 NC 3 2 1 -15i -15I 100nF -15i Figure C.2: VoltageSensor.SCHDOC [38] 132 C.1. Schematics of Control Board -15i 3 2 1 R1 Res1 62 CN2 +15i R2 R3 Res1 32K C27 30K 150pF +15i C28 4 4 LM324AM U1A C29 11 2 1 1 R6 12K 2 6 5 U2 LM4040A30IDBZR-3.0 7 2 11 2 3 +15i 3.48K +15I 100nF 4 R5 5.1K R4 3 LM324AM U1B -15I 1 1 11 -15i CM1 LM324AM U3A -15i NC 100nF 1 3 -15i -15i 3 2 1 R7 Res1 62 CN3 +15i 30K C30 150pF R10 3.48K +15I 8 R11 LM324AM U1C 12K 11 3 -15I 6 13 12 14 4 LM324AM U1D 11 9 10 +15i 4 4 +15I 4 R8 Res1 32K 5 7 2 CM2 LM324AM U3B 11 R9 -15i -15I Figure C.3: CurrentSensor.SCHDOC [38] 133 C.1. Schematics of Control Board +5 CN4 4 3 2 1 +15 +5 TMP CN5 2 1 5K R12 3K 1.5K TMP9 TMP9 12 R13 R14 5 CM1 U3A 2 TMP CM2 4 3 LM339AM C31 2.5v D4 300 R15 +5 100nF 1 3 5 7 9 11 13 15 17 19 2 4 6 8 10 12 14 16 18 20 VM1 VM2 VM3 FRQCN SW JP1 SW CN6 4 3 2 1 Header 4 +5 FRQCN Figure C.4: TMPSensor.SCHDOC [38] 134 C.2 Schematics of Power Board +5 P1 TMP P2 4 3 2 1 2 1 TMP9 Header 2 +15 Header 4 C1 10uF C2 0.1uF U1A 5 4 6 2 1 +5 6 D11 U2A 1 TMP 15V D12 SN74LVC2GU04 +15 Concept1 SN74ALS21AD 4V7 D2 R2 4.7K +15 JP1 40 38 36 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 +15 39 37 35 33 31 29 27 25 23 21 19 17 15 13 11 9 7 5 3 1 UC2 +5 +15 R3 22K R4 22K +15 +15 +5 +15 150pF C4 150pF R5 4.7K R1 4.7K D3 4V7 C3 R6 22K R722K C5 150pF C6 150pF R84.7K R9 4.7K D4 4V7 +15 R10 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 GND W_SO1 W_VL W_RC1 W_InA W_InB W_RC2 W_Mod W_SO2 VDD VDD GND GND V_SO1 V_VL V_RC1 V_InA V_InB V_RC2 V_Mod V_SO2 VDC VDC GND GND U_SO1 U_VL U_RC1 U_InA U_InB U_RC2 U_Mod U_SO2 GND 6SD106E C7150pF 22K +15 R11 68 67 66 65 64 63 62 61 60 59 58 57 56 55 54 53 52 51 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 D5 R13 39K R14 180 15V D13 UE2 UG2 15V D14 UC1 R15 D6 39K R16 180 UE1 VC2 R17 R18 D7 39K 180 15V D15 UG1 15V D16 VE2 VG2 15V D17 VC1 R19 R20 D8 39K 180 VE1 WC2 R21 R22 D9 39K 180 15V D18 VG1 15V D19 WE2 WG2 15V D20 WC1 R23 R24 D10 39K 180 WE1 15V D21 WG1 15V D22 15V C8150pF 22K +5 U_C2 U_Rth2 U_E2 U_G2 Free Free U_C1 U_Rth1 U_E1 U_G1 Free Free V_C2 V_Rth2 V_E2 V_G2 Free Free V_C1 V_Rth1 V_E1 V_G1 Free Free W_C2 W_Rth2 W_E2 W_G2 Free Free W_C1 W_Rth1 W_E1 W_G1 R12 4.7K 135 Figure C.5: Driver.SCHDOC [38] C.2. Schematics of Power Board 14 7 D1 16V C.2. Schematics of Power Board C9 * 2700uF C10 R25 * Thermistor 5 1.5uF C11 * 1.5uF UC2 UE2 UG2 UC1 UE1 UG1 VC2 VE2 VG2 VC1 VE1 VG1 Inv1 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 3 2 1 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 3 2 1 4 5 6 7 8 9 4 5 6 7 8 9 1 2 3 JP3 +5 TMP9 MUBW20-06A7 WC2 WE2 WG2 WC1 JP2 1 2 3 WE1 WG1 Figure C.6: MUBW20.SCHDOC [38] 136 C.3. PCB Layouts C.3 PCB Layouts Figure C.7: Control board PCB [38] 137 C.3. PCB Layouts Figure C.8: Control board PCB top layer [38] 138 C.3. PCB Layouts Figure C.9: Control board PCB bottom layer [38] 139 C.3. PCB Layouts Figure C.10: Power board PCB [38] 140 C.3. PCB Layouts Figure C.11: Power board PCB top layer [38] 141 C.3. PCB Layouts Figure C.12: Power board PCB bottom layer [38] 142 Appendix D Schematics for 3-Phase Islanding Test 143 Appendix D. Schematics for 3-Phase Islanding Test Figure D.1: Schematics for 3-phase islanding test [38] 144 Appendix E Unintentional islanding test conditions Unintentional Islanding Test Test conditions: Single Phase 120V /60H Z /1kW 1. PF=[ 1, 0.37, 0.707 ] ⇔ Q f =[ 0, 2.5, 1 ] 2. Values Table: Criteria Value Unit Pload 1000 W Qload 0 VAR Matched LC VEPS 120 V RMS fEPS 60 Hz VIUT 120 V PIUT 1000 W IIUT 8.33 A PFIUT 0.95 =18.195o Qf 1 RLC below designed for this Qf Rload 14.4 OHM LLoad 38.197 mH R / 2 f o Q f , iL(0)=11.785A CLoad 184.207 F Q f / 2 f o R Qf 2.5 Rload 14.4 OHM LLoad 15.28 mH R / 2 f o Q f , iL(0)=11.785A CLoad 460.52 F Q f / 2 f o R Q f =R Notes Inverter Under Test RMS V2/P RLC below designed for this Qf V2/P 1 C for a parallel RLC load, and Q F = −1 . L PF 2 P IUT =P LoadP EPS & QIUT =Q Load QEPS 145