This action might not be possible to undo. Are you sure you want to continue?

Dissertation submitted to the College of Engineering and Mineral Resources at West Virginia University in partial fulﬁllment of the requirements for the degree of Doctor of Philosophy in Electrical Engineering

Professor Ian Christie, Ph.D. Professor Ali Feliachi, Ph.D. Professor Powsiri Klinkhachorn, Ph.D. Professor Muhammad A. Choudhry, Ph.D., Co-Chair Professor Ronald L. Klein, Ph.D., Chair Lane Department of Computer Science and Electrical Engineering

Morgantown, West Virginia 2002

Keywords: distributed generations, dynamic stability, control, power systems Copyright 2002 Zhixin Miao

Abstract

Modeling and Dynamic Stability of Distributed Generations by Zhixin Miao Doctor of Philosophy in Electrical Engineering West Virginia University Professor Muhammad A. Choudhry, Ph.D., Co-Chair Professor Ronald L. Klein, Ph.D., Chair

The objective of this dissertation is to develop dynamic models for distributed generations (DG), to investigate their impacts on dynamic stability of power distribution systems, and to design controllers for DGs to improve the dynamic stability of the integrated power distribution system. A two-year distributed generation (DG) project at West Virginia University (WVU) evaluated the impact of various DG sources on actual distribution systems by performing computer simulations. The data is supplied by two regional electric utilities of two actual distribution systems each. In this project several important issues were investigated, including the availability of simulation tools and impacts of DGs connected to a distribution line under a variety of line operating conditions. Based on this preliminary research the further most interesting topics for continued research were raised. The continued research has focused on deeper investigation, such as, modeling DG sources, evaluating their interaction and impacts, and improving the dynamic stability of the integrated power distribution system. Four speciﬁc DGs are studied in this dissertation: fuel cell power plant, wind turbine induction generator, gas turbine synchronous generator and diesel engine synchronous generator. A full-order synchronous generator model represents the generator models of gas turbine generator and diesel engine generator. A simpliﬁed gas turbine model has been chosen to be implemented. A practical diesel engine for emergency use is modeled. The generator model of wind turbine induction generator is represented by a full-order induction generator. The rated power operating regime is considered for impacts evaluations and controller design. Two types of fuel cell models are developed. The ﬁrst one is a model of already operational phosphoric acid fuel cell (PAFC) obtained through data ﬁtting and the second one is dynamic model of solid oxide fuel cell (SOFC). Since fuel cells are connected to the electric power network via inverters, an inverter model has been developed. Multi-DG controls are investigated in this dissertation. One DG control is fuel cell control, the other one is wind-turbine control. The control of fuel cell (SOFC) plant is through the inverter to adjust active power injection to the network during the transient time. The control of wind

turbine generator is through the parallel connected SVC by adjusting reactive power injection to the system. Both control schemes are centralized. Linear analysis methodologies are utilized in designing the controller. In the fuel cell control design, two pairs of critical modes are screened out using eigenvalue analysis. The participation factors of DGs with respect to the modes are calculated. Two speciﬁc lead-lag compensation units are designed to damp each mode separately. The gains of the two compensation units were then obtained via optimal control methodology. In wind turbine DG control design procedure, three rotor speed deviations are used as input signals while the controller outputs are the ﬁring angle for the SVC and the pitch angle for the wind-turbine DG. An output feedback controller is designed. The dynamic load characteristic is also considered by modeling it as a structured uncertainty. µanalysis is used to evaluate the robust stability of the controllers with respect to the uncertain parameters in the dynamic loads. The IEEE-13 node radial feeder with existing gas turbine and diesel engine DGs is used as a test system to evaluate the multi-DG control. The simulation results demonstrate the eﬀectiveness of the control strategies. Coordinated operation of all the DGs is investigated. Simulation results show that good conﬁgurations within DGs along the system can improve the system stability. Furthermore, the fast acting SVC is very eﬀective in improving damping. Among the DGs investigated in this research, the fuel cell plant control is the best choice for the coordinated operation. Finally, the approach to model a complete three-phase power distribution system is implemented. The impact of the developed DGs models is evaluated on a three-phase unbalanced distribution system. The three-phase 13-node IEEE system with gas turbine and diesel engine DGs is simulated using MATLAB/Simulink’s Power System Blockset (PSB). In the simulation, a three-phase thyristor controlled braking resistor (TCBR) is connected to absorb the surplus energy when the system is subjected to a disturbance. The three-phase dynamic simulation demonstrates the eﬀectiveness of the proposed strategy.

My gratefulness goes to my parents. A special thanks goes to my wife for her love and supporting throughout my study. To Prof. program. . chair of my committee.iv Acknowledgements I am very glad to thank people who helped me during my study to pursue the Ph. for they raised me up and continually encourage me to pursue my career. Prof. Muhammad Choudhry. for his generous help and contribution on my academic progress. degree. Christie for serving in my committee and providing me helpful advice to my work besides their well-taught courses. Ronald Klein.D.D. Feliachi. I would like to thank Prof. for his continuous support over the years. co-chair of my committee. To Prof. Klink and Prof. his guidance on my research and his patience and kindness. His advice and help make it possible for me to ﬁnish the Ph.

. . . . . . . . . . . . . . . . . . . . . . . .2 Models . . . . . . . . . . . . .4 Coordinated Control . . . . . . . . . . iv x xiv 1 1 2 4 4 4 5 6 8 9 11 11 11 12 13 13 13 14 14 14 16 16 16 16 17 18 . . . . . . . . .3 Fuel Cell . . . . . . . . . . . . . . . 2. .4. . .1 Problem Description . . . . . . . . . . . . . . .2 Model Reduction . . . . . . 3.3. . . . . . . . . . 2. . . . . . . . . . . . . . . . . . . . . . . . .2 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. . . .1 Introduction . . . . . . . . . . . . . . . . . . . . . . .4 Wind Turbine . . . . . . . . . . . . . . . .4 Power Electronics Application . . . . 2. . . . . . . . . . . . . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Gas Turbine . . . .5 Robust Analysis of the Load Parameter Uncertainty 2. . . . . . . .2. . . . . . . . . . . . .2. . . . . . . . . . . . . . .1 Linear Analysis .3 Test Cases and Simulation Results .2 Diesel Engine . 1. . . . . . . . . . . . . . . . . . . . . . .3 Control Strategies for DGs . 2. . . . . . . . . . . . . .1 Static VAR Compensator (SVC) .2. . . . . .3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. . . . . . . . .3 Thyristor Controlled Braking Resistor (TCBR) . . . . . . . . . . . . . . . . . 3. .3. . . . 2. . . . . . 2. . . . . . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . .3.1 Introduction . . . . . . . .2. 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .v Contents Acknowledgements List of Figures List of Tables 1 Introduction 1. . .2 Models of An Actual Distribution System 3. . . . .2 Current/Voltage Source Inverter (CSI/VSI) . . .3. . . . . . . . . . . . . . . . . 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. . . 2. . .2. . . . . . . . . . . . . . . . . . . . . . . . . . 3. . . . . . . . . . . . . . . . . . . . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Control Schemes . . . 2. . . . . .2 Models of Distributed Generations . . . . . . . .1 Suburban System I (SI) . . . . . . . . . . . . 2. . . . . 2 Literature Survey 2. . . 3 Preliminary Research of DGs 3. . . . . . . .

. . . . .3 Fuel Cell Transient Output Voltage Modeling . . . . . . . . . . . . . . . . . . . . 4. .3 Gas Turbine . .5 Inverter Control Unit . . . . . . . . . . .6 Model Simulation and Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Diesel Engine .2 System Models . . . . . 6. . . . . . . . . . . . . . . . . . .4 Models Linearization . . . . . . . . . .4 Conﬁguration of Test System with DGs . . . . . .3. . . Summary . . . . 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. . . . . . . . . . . . . . . . . . . . . . .1 Introduction . . . . . 4. 5. . . . . . . . . . . . . . .2 Eigenvalue Sensitivity . . . . . . . . . . . . . . . . . . . . . . . .3. . . . .3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 Summary . vi 19 19 21 21 24 24 25 25 26 27 28 28 29 31 32 32 33 35 35 37 39 40 40 41 43 43 44 46 49 49 51 52 55 55 56 57 58 59 61 62 3. . . . . . . . . . . . . . . . .3 Data Fitting . . . . . . . . . . . . . . . . . . . . . . . . .5 Eigenvalue Analysis . . . . . . . .3 Dynamic Model of Solid Oxide Fuel Cell (SOFC) . .2. . . . .5 Simulation Study . . . . . . . . . . . . 4. . . . .2 V-I Characteristics . . . . . . . 6 Dynamic Modeling of A Solid Oxide Fuel Cell Power Plant 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. .1 Eigenvalue of the Distribution System with DGs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3. . . . . . . .4. . . . . . . . . . . . . 6. . . . . . . . . . . . . . . 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. . . . . . . . .2 Substation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Full-Order Synchronous Generator . . . . . . . 6. . . . . . . . . . . . . . . . . . . .2 Fuel Cell Transient Output Current Modeling . . . .2. . . . . . . .2. . . . .3 Reclosing Time .5. . . . . . . . . . . . . . . . . . . . . . . . . . . 5. . . . . 4. . . . . . . . . . .4. .6 Summary . . . . . . . . . . . . . . . . . .1 Introduction . 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Introduction . .4 4 DGs Simulation and Linear Analysis with Simulink 4. 3. . . . .7 Summary . . . . . . . . . . . . . . . 6. . . 5. . . . . . . . . . 4. . . . . . . . . . . . 4. . . . . . . . . . . . . . . . . . . . . . . . 5. . . . 6. . . . . . . . . . . . . . . . . . . . . . .3. . .2 Structure of A SOFC Power Plant . . . . . . . . . . . . . . . . . . . . . .CONTENTS 3. . . . . . . . .3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. . . . . . . . . .2 Linearized Gas Turbine Synchronous Generator . . . . . . . . . . . . .1 Fault Location . . . . . . 4. . . . . . . . . . . . . . . in Simulink . .3 Linearized Diesel Engine Synchronous Generator . . . . . . . . . . . . .5. . . . . . . . . 5 A Phosphoric Acid Fuel Cell Model from Data Fitting 5. . . . . . . . . . . . .4 Fuel Cell Modeling . . . 6.5 Average Model of IEEE-13 Node Feeder . . . . . . . . . . . . . 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2. . . . . . . 5. . . . . . . . . . . . . . . . . . . . . . . . . . . .2. . . . . . . . . . . . . . . . .3 Building Synchronous Generator Model using Matrix Concept 4. . . . . . . . 5. . . . . . . . . . 4. . . . . . . . . . . . . . . . . . . . . .4. . . .3. . . . . . . . . . . . . . . . . . . .4 Power Conditioning Unit (PCU) . . . . . . .1 Linearized Substation Model . .1 Least Squares Optimization Algorithm . . .2 Fault Clearing Time 3. . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .5 Integration of Wind Turbine DG . . . . . . . . . . 8. . . . . . .4 Controller Design . . . 7. 9. . . . . . . . 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . 9. . . . . . . . . . . . . . . . . . . . . . . 9. 8. . . . . . . . . .4. . . . . . . . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7. . . . . . . . . 8. . . . . . . . . .4 Static and Dynamic Load Models . . . . . . . . . . . . . . . . . . . . . . . . . . . .4. . . . . . . . .3 Analyzing Robust Stability . . . . . . . . . . . . . . . . . .3 System Linearization .3. . . . 9 Robust Control of Wind Turbine DG 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 System Conﬁguration . 9. .6. . . 8. . . . . . . . . . . . . . . . . . . . . . . . . . . 9. . . . . . . . .1 Introduction . . .2 Optimal Control Algorithm to Select Gains . . . . . . . . . . . .4 Induction Machine Simulation . . . . . . . . . . .2 Dominant Eigenvalues Identiﬁcation . . .2. vii 65 65 65 66 69 71 72 73 74 79 79 80 80 81 81 82 83 84 85 85 86 88 88 90 92 94 94 94 95 95 97 98 98 98 99 102 102 103 104 106 108 . . . . . . .3. . . . . . . . . . . . . . 8. . . . . . . . . . . . . . . . . . . .2 Induction Machine Simulink Model . 8 Model of the Wind Turbine DG 8. . . . . . . . . . . . . . . . . .4. . . . . . . . . . . . . .5 Linearized Induction Machine . . . . .2. . . .4. 8. . . . . . . 9. . . . . . . . . . . . . . . . . . . . . . .1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6.2 Induction Generator . . . 9. . . .1 Multivariable Coordinated Controller 9. .2 System Models . . . . . . . . . . . . . . . . . .1 Matrix Concepts Based Induction Machine Model . . . . . . . . . .7 Simulation Results of Robustness Analysis . . . . . . . . . .8 Summary . . . . . . . . . . . . .4. . 9. . . . . . . . . . . . . . . . 8. . . . . . . .6 Controller Robustness Analysis . . . . . .3 Control Unit Model . . . . . . . . . . . . . . . . . . . . . . . . .6 Summary . . . .5 Simulation Results . . . . . . . . . . . . . . . . . . . .6 Summary . . . . . . . . . . . . . . . . . . .CONTENTS 7 Dynamic Stability Control for A SOFC Power Plant 7. 9. . . . . . . . . . . . . . . . . . . . 8. . . 7. . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Initialization of the Induction Machine . . . .2 System Conﬁguration . . . . . . . . . . . .2 Design Results . . . . . . . . .4 Controller Design to Assign Eigenvalues . . . . . . . . 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Critical Eigenvalues Identiﬁcation . . . . . . 9. . . 9. .6. . . . . . . . . . . .4. .3 Static VAR Compensator (SVC) . .1 Phase Angle Compensation . .2 Parametric Uncertainty . . . . . . . . . . . . . . . . .4 Start-up Procedure and Step Response of Induction Machine 8. .5 Nonlinear Simulation to Test Controller . . . . . . . . . . . . . . .1 Introduction . . . . . . . . . . . . . 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4. . . . 7. . . . . . . . . . 8. . . . . . . . . . . . . . . . . .1 Wind Turbine Model . . . . . .1 Representing Uncertainty . . . . . . . . . . . . . .4. . . . . . . 8. 8. . . . . . . . . . . . . . 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7. . . . . . . . . . . . . . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . .1 Ranking Indices of the Eigenvalues . . . . . . . . . . . 8. . .

. . . . . . . . . . . . . . . . . 118 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2. . . . . . . . . . . .5 Parameters for TCBR Simulation in Chapter 11 C Program Structure of C. . . . . C. . . . . . . . . . . .1 Parameters for SI system in Chapter 3 . . . . . . . .3. . . . . . . . . . . . . . . . . . . . . A. . . . . . . . . . . B. . .4 Test System of TCBR . . . . A. . . . . . . . . . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . .5 Dynamic Simulation Results . . . . . . . . . . . . . . . . . . 111 . . . .2 Parameters for the Test System in Chapter 4 . . . 115 . 11. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. . 11. . . . . . . . . . . . .2 System Models . . . . . .1 Basic Concepts of Variable Structure Controller . . 11. . . . . . . 11. . . . . . . . . . . . . . Simulation . . . . . . . . . . . . . . . . . .4 Voltage Source Inverter . . . . . . . . . . . DGs . . . . . . . . . . . . . . .3. . . . . . 11. . . . . . . 11. . .5. . . . . . C. . . . . . . . . .1 Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Parameters for the SOFC Simulation . . . .1 Excitation System Model . . . . . . . . . . . . . . B. . . . . . . . . . . . . . . . . . . . . . . A. . . .6 Summary . . . . . .4 Impacts of Diﬀerent Real Power Penetrations 10. . . . . . . . .2 Interaction of DGs . . . . . . . 111 . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Analysis Tools . . . . 10. . . . . . . . . .5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .CONTENTS 10 Coordinated Operation of Multiple Types of 10. .3. . . . . . . . . . . . . . . . 11. . . . . . . . . . . .5 Test Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 Dynamic Braking Control Strategy . . . . . . . . . . . . viii 111 . . . . . . . . . . . . . . . . 10. . . . . . . . . .2 Line Segment Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. . 11. . . . . . . . . . . . . . . . . . . . . . .3 Critical Eigenvalues and Controller Design .2 Thyristor Controlled Braking Resistor (TCBR) . . 119 120 120 121 121 121 123 123 123 124 125 127 127 130 133 143 143 146 147 149 149 150 150 154 154 154 155 155 155 157 157 159 161 161 . . . . . . . . . . . . . . . . . . . . . . . . . .3 Three-Phase Control Strategy . . . . .4 Features .5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 IEEE-13 Node Test Feeder . . . . . . . . . . 11 Three-Phase Distribution System Study 11.3 Induction Machine . . . . . . . . . . . . .2 Simpliﬁed Synchronous Generator Model . . . . B Parameters for Simulation B. . . . . . . . . . . . . . . A. B. . . . . .3 Controller Design . . . B. . .1 Full-Order Synchronous Generator Model A. . . . . . . . .1 Introduction . . . 11. . . . . . . . . . . . . . . . . . . . . . . .4 Parameters for the Wind Turbine DG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. . . . . . . . . . 10. . . . . . . . . . . . . . . . . . .2 Network . . . . . . . . . . . . . . . . . . . .1 Introduction . . . 12 Conclusion References A Conﬁguration of Models A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

CONTENTS D Robust Analysis Program Vita ix 163 165 .

. . . .5 3. . . . 22 .3 3. . . . .6 5. . .x List of Figures 2. . . . 19 .2 3. . . . . . . An equivalent circuit of the PAFC. . . . . . . . . . . . . .7 4. . . . .3 3. . . . . . . . . . . Y axis: voltage (V)) . . . . . . . . . . Using matrix/vector to represent the synchronous generator model in Simulink. . . . . . . 6 Structure of a fuel cell power plant. . . .1 3. . . polynomial order for the PAFC’s DC voltage. . .2 5. . . 41 . . . . . . . .9 General structure of a diesel-engine DG. . . . 22 . Diagram of the substation model of the distribution system. . . . .8 5. . . . . The recorded data and ﬁtted curve of the PAFC’s DC current (Case 2). . . . . . Linearized diesel engine powered synchronous generator. . . . . . . . . . . .8 5. . . . . . . . . . . . . case 22). . . . y axis: voltage (v)). .2 4. . . . . . . . . . (x axis: time (s). . . . 41 . . . . . . . . . . 17 . . . . . . . . . . . Linearized gas-turbine powered synchronous generator. . . . . . . . . . . . . . . case 25). . . 48 . . . IEEE-13 node distribution system with a gas turbine DG and a diesel engine DG. . . . . . Diagram of a diesel engine and a governor model. . . . . . . . . . . . Diagram of the linearized substation model of the distribution system. . . . . . . . . . . . .4 4. . . . 9 A wind-synchronous generator with a control loop.5 5. . . . 21 . . . . . . . . .6 4. . . . case 22. . .5 4. . . . . . . The voltage magnitude of the substation with diﬀerent fault location (case 20. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Diagram of the Error vs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 2. . . . . . . . Diagram illustrating the procedure for representing three-phase as a single-phase equivalent circuit. . .6 4. . The recorded data and ﬁtted curve of the PAFC’s DC current (Case 1). . . . . . . .1 5. . . The recorded data and ﬁtted curve of the PAFC’s DC voltage (case 1). . . . . . A model of the PAFC’s DC part. I-V characteristics of the PAFC obtained from the recorded data. . .4 3. . . . . . . . . . . . . . . . . . . case 30). . 26 28 28 29 31 32 33 34 . 10 One-line diagram of the SI distribution system with DGs. . . . . .3 5. . . . Diagram of the sampled data of the DFR device. . . . . . . . . . . . . Diagram of a PAFC’s generalized schematic. . . . . . . . . .1 2. . .4 5. . . (X axis: current (A). . . . . . 18 . . . . . . . . . . . . . . . . . Frequency of DG1 with diﬀerent DG1’s location (case 22. . .7 5. . . . . . . . . . . . . . . . . . . . . .1 4. . . . Diagram illustrating the procedure used to combine multiple lines and multiple loads into an equivalent single line and single load equivalent circuit. . . . . . . . . . . . .3 4. . . . . . . . . . . . . . . Diagram of a simpliﬁed gas turbine model. . 42 42 43 45 45 48 . . . . . . . . . . . . . . Frequency of DG1 with diﬀerent fault location (case 9. . . . . .

Terminal voltage of the SOFC when a fault occurs at bus 632 (Case 3). . . . . . . . . . . .17 Rotor speed response of the diesel-engine synchronous generator during the PAFC’s turn-oﬀ procedure (dotted line: Case I. . . . . . 54 . . . 67 . Three-phase CSI-based PCU. . . . . . .7 6. . . . . . . .8 7.LIST OF FIGURES 5. . . 52 . . . . Diagram of the current control scheme for the CSI-based PCU. . . . . . . . . . . 68 69 71 74 75 75 76 76 77 77 78 78 . . . . . . . . . . . . . . . Y: voltage (V)). . . . . . IEEE-13 node power distribution system with multi-types of DGs to evaluate the impact of a SOFC power plant. . . . . . . . . .2 7. . .14 7. . . . . . . . . . .8 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 6. . . . . . . . solid line: Case II). . (X axis: time (s). . .6 7. . . . Reactive power response of the step change in real power set point. . . . . . Y axis: impedance (Ω)). . . .11 The V-I curve of the obtained PAFC’s model. . . . . . . . . . . . . . . . . . . . . .14 Real power penetration from the PAFC power plant to IEEE-13 node power distribution system (dotted line: Case I. . . 5. xi . . . . . . . . .9 7. . . . . . . . Structure of a fuel cell. . . 51 . . . . . SOFC internal Nernst’s voltage response of the step change in real power set point. . . . . . 66 . . Rotor speed of the gas turbine DG when a fault occurs at bus 632 (Case 2). . . . . . . . Rotor speed of the diesel engine DG when a fault occurs at bus 632 (Case 3). .15 Voltage magnitude of bus 671 during the PAFC’s turn-oﬀ procedure (dotted line: Case I. . . . . . . 5. . . . . . .12 7. . . .9 6. Terminal voltage of the SOFC when a fault occurs at bus 684 (Case 2). . . . . (X: current (A). . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 7. . Rotor speed of the gas turbine DG when a fault occurs at bus 634 (Case 1). 5. . . . . . . . . .10 Diagram of the proposed PAFC’s dynamic model. 67 . . SOFC terminal voltage response of the step change in the real power set point. . . . . . time.1 7. . 5. 6. .10 7. . . . . . . . . . . . . . . . . . Terminal voltage of the SOFC when a fault occurs at bus 634 (Case 1). . . . Rotor speed of the gas turbine DG when a fault occurs at bus 684 (Case 2).5 7. . . . . . . . . . . . . Diagram of an average model of a CSI-based PCU. . Structure of the linearized output feedback control. . . . 53 . . . . Terminal and internal Nernst voltage of SOFC as a fault occures at bus 632 without controller. . . . . . . . . . .3 7. . . . . . . . . 50 . . . . . . .1 6. . . . .10 7. . . 56 57 58 60 61 62 63 63 64 64 . . . . . . . . . . . . . 49 . Real and reactive power of the SOFC power plant as a fault occures at bus 632 without controller. . solid line: Case II). . . . . . . . . . . Rotor speed of the diesel engine DG when a fault occurs at bus 684 (Case 2). . . . . . . . .15 Diagram of the general structure of a SOFC power plant. . . .4 7. . . . . . .13 7. . Rotor speeds of DGs 1&2 as a fault occures at bus 632 without controller. . . . . Diagram of single fuel cell power plant and inﬁnite bus. . . . . . .12 The impedance of the obtained PAFC’s model vs. . . . . 50 . . . . . . . . .3 6. . . . . Real power response of the step change in real power set point. . . . . . . . . . . . 53 . . .11 7. . . . . . . . . . . . . 5. . .6 6. . . . . . . Coordinated control scheme of the SOFC power plant. . . .13 IEEE-13 node power distribution system with mult-types of DGs. . . . . .2 6. . . . . . Rotor speed of the diesel engine DG when a fault occurs at bus 634 (Case 1). . . . . . . solid line: Case II). solid line: Case II). . .4 6. . . 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . .16 Rotor speed response of the gas-turbine synchronous generator during the PAFC’s turn-oﬀ procedure (dotted line: Case I. . 5. . . . .

. . . . (X axis: frequency (rad/sec)) . . . . .005 increase. . . . . . . . . .2 9. . . Diagram of the SVC (FC-TCR) structure. . . . . . . . . . Diagram of the blade pitch angle controller of the wind turbine. . . .8 9. . .4 9. . . . . . .9 8. . . ∗ Real and reactive power responses of the wind turbine DG to track Pwind (Case I). . . . . . . . Terminal voltage of the wind turbine DG to track the Q∗ wind Rotor speed of the wind turbine DG to track the Q∗ wind (Case II). . . . . . . . . Real/reactive power responses of the wind turbine DG to track the Q∗ wind (Case II). . . . . . . . . . . . . . . . . . . . Rotor speed of an induction machine during the start-up and step response procedure. . . . . . . . . State equations of an induction machine represented in Simulink. . . . .3 8. . . . . . . . . .3 9. .15 9. . xii 80 82 83 84 84 86 86 87 88 88 89 89 91 91 92 93 93 95 100 100 101 101 102 103 104 105 106 107 107 108 109 109 110 9. . . . . . . . . . µ upper bound and lower bound of system with the three parameters (rs . . . an induction generator and a SVC. . . (Case II). . . . . Diagram of the susceptance of SVC with controller.14 8. . . . . . . . . . State variables of induction machine during the start-up and step response procedure. . Susceptance of SVC with a controller when a fault occurs at bus 632. . . . . . . . . . 112 . . .11 9. . . . . . . . . . .9 9. . . .11 8. . . . . . Swing equation of an induction machine represented in Simulink. . . . . . . . . . . . . . . . . . . . . . .14 9. .10 9. . Electrical torque response of an induction machine during the start-up and step response procedure. . . . . . . . . . . . . . . XM ) of dynamic loads increasing 0. Diagram of the interconnected system. . . . . . . . . . . .02 increase. . . . . Diagram of a one-machine inﬁnite bus system to test the induction generator. . . . . . Digram of the feedback loop. . . . . . . . . . . . . . .12 9. . . . . . . . . . . . . . . . . . rr . . . .13 8. .1 8. . .1 Diagram of IEEE-13 node power distribution system with multi-type of DGs to investigate the coordinated operation. . . . . . . Rotor speed of the wind turbine DG with modiﬁed parameters of dynamic loads when a fault occurs at bus 632 (dotted line: without controller. . . . . . . Rotor speed of the wind turbine DG with/without a controller of the SVC. . . . . . . . . . . . . . . . Diagram of the SVC’s voltage regulator control loop. . . . . . . . . . . . . . . . . . . . . . . . . . Diagram of Mu (µ) as rs_mot has a 0.12 8. Rotor speed of the gas turbine DG with modiﬁed parameters of dynamic loads when a fault occurs at bus 632 (dotted line: without controller. . .10 8. . . . . . . . Rotor speed of the diesel engine DG with/without a controller of the SVC. . . . . . . . . . .LIST OF FIGURES 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 8. . . . . . . . . . . Diagram of Mu (µ) as rs_mot has a 0. . . . . . . . . . . .1 9. . . . . . . Frequency response of M11 . solid line: with controller). General structure of a wind turbine DG and its control unit. Rotor speed of the gas turbine DG with/without a controller of the SVC. . .5 8. . . . . . . . . . . . . . . . . . . . . . . . . . ∗ Terminal voltage response of the wind turbine DG to track Pwind (Case I). . . . . . . . Rotor speed of the diesel engine DG with modiﬁed parameters of dynamic loads when a fault occurs at bus 632 (dotted line: without controller. . . . Equivalent circuit of an induction machine. solid line: with controller). . . . . . . .6 9. . . . . . . . . . .13 Diagram of a wind turbine DG consisting of a wind turbine. . Diagram of the linear fraction transformation. . . . . . . . . . .7 8. . . . . . . . . . . . . . . . . . . .5 9. solid line: with controller). .7 9. . . . .4 8. .02pu. . . . .8 8. .17 9. . . . . . . . . . . . . .15 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16 10. . . . . . .16 8. . . Diagram of IEEE-13 node distribution system with multi-DGs to evaluate the wind turbine DG. . . .2 8.

. . . . . . . . . .3 A. . . . C. .5 A. . . Diagram of a simpliﬁed synchronous generator model. . . .3 11. . . .LIST OF FIGURES 10. . 159 . . . . . . . . . . . . . . .9 Rotor Rotor Rotor Rotor Rotor Rotor speed of the gas turbine DG with/without SOFC connected. .1 11. . . Zero-axis equivalent circuit of an induction machine. . . . Terminal voltage of DG2 as a fault occurs at bus 684 with/without TCBR. . . . . . . . . . . Rotor speed deviation of DG1 as a fault occurs at bus 684 with/without TCBR.2 General structure of the simulation program in Simulink. . .1 General structure of the simulation system. . . .4 11. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 C. . . . . . . . . . . . . . . . D-Axis equivalent circuit of an induction machine. . . . .5 11. . . . . . . . . . . . . . . . . . system. . . . . . . . . . . . . . .3 Simulation of a SOFC in Simulink. . . . . .6 A. . . . .8 A. Diagram of IEEE-13 node distribution system with multi-DG and TCBR. . . . . . . .2 A.2 10. . . Terminal voltage of DG1 as a fault occurs at bus 684 with/without TCBR. Diagram of the three-phase TCBR and its control unit. . . .7 11. . . . . speed of the wind turbine DG with/without coordinated controller. .6 11. . .1 A. . . . . . . . . . . . .4 A. . . . . . . speed of diesel engine DG with/without coordinated controller. . . . . . . . . . . . . . . . Q-axis equivalent circuit of a synchronous generator. 158 C. . Q-Axis equivalent circuit of an induction machine. . . . . . . . . . . . . speeds of the gas turbine DG with diﬀerent controllers. speed of the wind turbine DG with/without SOFC connected. . . . . . . . . Zero-axis equivalent circuit of a synchronous generator. . . . . . . . . . . . . . .3 10. . .6 10. . . . . . . . . . Stator speed deviation of DG2 as a fault occurs at bus 684 with/without TCBR. . . . . . . . . . . . . .5 10.2 11. . . .7 A. . . . . . . . Equivalent circuit of one phase TCBR. . . . . . . . D-axis equivalent circuit of a synchronous generator. . . xiii 113 114 114 116 117 117 121 122 125 126 127 128 128 129 144 145 145 146 147 147 148 149 150 Diagram of the excitation system. . . . . . . . . . . . . Diagram of a PWM current source inverter in a distribution One line diagram of IEEE-13 node radial feeder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8 A. . . . . . . . . . . . . . . . . . . . . speed of the diesel engine DG with/without SOFC connected. . .7 11. . . . . . . . . . .4 10. . .

. . 113 Critical Eigenvalues of the open linearized system with multi-DGs. . . . .1 7. . . . . .4 4. . . . . . . . . . . . . . . . . . . .2 Three-phase current and voltage of distribution system without DG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 4. . . . . . . . .2 10. Dominant eigenvalues of the open/closed loop system. . . . . . . . . . . . . . .4 10. . . . . . . .3 4. . . . . polynomial order. . . . . . . . . . DGs. . . . 155 . .3 7. . . . . . . 90 Eigenvalues of the open-loop linearized system.4 8. 47 Dominant eigenvalues of the linearized distribution system Participation factors of eigenvalues λ21. . . . . . . . . . 151 A. .5 5. . . 69 70 73 73 . . . 118 Critical eigenvalues of the open linearized systems with diﬀerent real power penetration. . . . . . . . . 46 Range of parameters and representative parameter values. . . . . . . . . . 46 Error vs.xiv List of Tables 3.1 5.118 Critical Eigenvalues of the multi-DGs system with diﬀerent controllers. . . . . . . .2 10. . . . . . . . . . . 153 B. .1 9. .1 Line segment data of IEEE 13 node test feeder. . . . . . . . 119 11. . 96 Critical eigenvalues of the open linearized system. . . . . . 126 A. . 20 . . .2 4. .3 7. . . . . . . . . . . . . . . . . . . . . . . . 115 Critical Eigenvalues of the multi-DGs system with diﬀerent controllers.1 9. . . . . . . .1 Parameters for the test system simulation. . record data. . . . . . . . . . . . . . . . . .3 10. . . . . . Eigenvalues of the linearized diesel engine DG. . . with . . . . . .1 4. . . Eigenvalue sensitivity with respect to Xd . . . . . . . . . . . . . . . λ25. . . . . . . . . . . . . . . . . . . . . . . . . .6 The tests on the SI distribution system. . . . DC current parameters vs. Eigenvalues of the linearized distribution system with DGs. . . . .2 5. . . . . . . . . . . . .22 . . . . . Eigenvalues of the linearized induction machine. Feedback gains through the optimal control algorithm. . . .2 7. . . . . . . . . . . . . . . . . . . . . . . . . . .1 Three-phase current and voltage of distribution system with DG.5 10. . . 34 35 35 36 38 Eigenvalues of the linearized synchronous generator with gas-turbine. . . . . . .26 . . . . . Eigenvalues of the linearized synchronous generator connected to diesel engine. . . . . . . . . . . . . . . . . . . . 3 . . . . . . . . . . . . . . . . . . . . 97 Eigenvalues of the linearized coordinated system. . . . . . .1 10. . . . . . . . 116 Diﬀerent real power penetration of the multi-DGs. . . . . . . . . . .

LIST OF TABLES

xv

B.2 Parameters for the SOFC simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 B.3 Parameters of the wind-turbine DG. . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 B.4 Parameters for the TCBR simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . 156

1

Chapter 1

Introduction

1.1 Problem Description

Distributed Generation (DG) has entered the vocabulary of global energy companies and has recently become a mainstream topic in energy planning, policy and new business ventures. In a restructured energy industry, DG technologies can help power providers capture new markets, serve high value customers, reduce infrastructure investment, and optimize asset utilization. Energy suppliers can gain and maintain competitive advantage with DG technologies, which include gas turbines, diesel engines, fuel cells, and wind turbines. Emerging options are being ﬁeld tested and deployed by early adopters who view DG as a dynamic opportunity to strategically transform the market and gain a competitive advantage with advanced technology. Penetrating the current market will depend on successful DG applications and promoting the technology as one part of a reliable, standardized and clear solution. Traditionally, power systems have been operated in a vertically structure. Almost all of the power has been generated at large power plants and the power is transmitted to consumers via long distance transmission system. Currently, substations are the only electric power source for loads in the distribution systems. With the advent of DGs, the newly power sources come out. Each of the DGs has its own advantages and disadvantages. While DGs are connected to the distribution system and to supply power, their connections will have a great impact on the real time system operation and many characteristics of their integration into existing systems are still uncertain, for example, the disturbance arising from fault or load ﬂuctuation could cause the DGs to fall out of step. The power distribution system has several speciﬁc features when compared to the power

CHAPTER 1. INTRODUCTION

2

transmission system: 1) Distribution systems are normally three-phase unbalanced due to the various customer loads; 2) Distribution systems are directly connected to the relatively large capacity transmission systems; 3) The lower voltage level results in higher resistor/reactance ratios (R/X) of the distribution lines. This R/X ratio will decrease when transformer and voltage regulators are considered. In this dissertation, dynamic stability controls are designed to improve the oscillation damping of the distribution system with DGs. To investigate the impacts of DGs on the dynamic stability of the distribution system, the four types of DGs: gas turbine synchronous generator, diesel engine synchronous generator, fuel cell power plant and wind turbine induction generator are studied. The dynamic models are expressed in state space and simulated using Matlab/Simulink. Dynamic stability controls of the distribution system with DGs are designed to improve the oscillation damping. The Flexible AC Transmission Systems (FACTS) devices such as: static var compensator (SVC), voltage source inverter (VSI), and thyristor controlled braking resistor (TCBR) are also studied. The coordinated control of DG and FACTS devices to improve the dynamic stability is an important topic in this research and has been addressed in this work.

1.2

Outline

The outline of this dissertation is summarized in the following. A literature survey is presented in Chapter 2 and gives an overview of state-of-the-art and world-wide DGs research. Various DGs are introduced and compared for their modeling development in the following chapters. Since dynamic stability control of power distribution system with DGs is a new emerging topic, papers on control theories and control application in power transmission system are surveyed. The literatures will provide ideas and justiﬁcation to select control schemes and device parameters for DGs’ dynamic stability control. The preliminary DG research on four speciﬁc power distribution systems is presented in Chapter 3. The power system toolbox (PST) is used to evaluate the impacts of DGs on four systems. The synchronous generator and induction machine models in PST are used to represent the DGs. This preliminary research provides the fundamental to deepen the DG modeling and dynamic stability research. In Chapter 4, the gas turbine and diesel engine DG models are presented. To simulate the full order synchronous generator, the matrix concept is used to simplify the model building process. A substation model for dynamic study is also presented.

In Chapter 11. Two types of fuel cells are studied: phosphoric acid fuel cell (PAFC) and solid oxide fuel cell (SOFC). The wind turbine DG modeling and control are studied in Chapters 8 & 9. To design damping controller for SOFC power plants.CHAPTER 1. The IEEE-13 node system with the gas turbine and diesel engine DGs is the test system to study the fuel cell and the wind turbine DGs. The robustness of the controller is investigated at diﬀerent real power penetration level. The robustness of the controller is veriﬁed over the parameter uncertainty of the dynamic loads. the three-phase approach is presented to evaluate the DG models on unbalanced distribution systems. A three-phase TCBR is connected to absorb the surplus energy when the system is subjected to a disturbance. the lead-lag units are used to compensate the dominant eigenvalues and an optimal control algorithm is applied to select the gains of the compensation units simultaneously. A coordinated operation of the multi-DG system is evaluated in Chapter 10. Chapters 6 & 7 focus on the dynamic model and control of a SOFC power plant. The controller is designed for a coordinated operation. The linearized model and eigenvalue based method are studied to design the dynamic stability controller. INTRODUCTION 3 Chapters 5-7 focus on fuel cells and their dynamic controls. An average power conditioning unit (PCU) is developed to connect the SOFC to the system. In Chapter 5. a PAFC model is obtained through the data ﬁtting method. The interactions between the DGs are studied. A multivariable controller is designed to control the wind turbine generator and SVC. . The conclusions and suggestions for future work are presented in Chapter 12.

However. photovoltaics and fuel cell will likely replace some conventional power generation [59]. Gas turbine. even simpliﬁed models covering dynamic characteristics of these types of DGs (and other newer DG sources such as photovoltaic and microturbines. 2. such as gas turbine.) are not openly available. wind turbine. the three-category of literatures are surveyed: distributed generation (DG) models. . modular installation. to evaluate their impacts on the dynamic stability of distribution systems and to design controllers to improve the dynamic performance of system. small scale power generation technologies. diesel engine. fuel cell and wind turbine DGs are four of the several types of new DG sources that have experienced considerable development progress in recent years.2 Models of Distributed Generations DG technologies oﬀer many advantages including high eﬃciency. diesel engine. Simulation of various types of DGs in a suitable software environment is the ﬁrst step in analyzing the dynamic characteristics of DGs and designing the control strategies. power electronics application and power system control schemes.4 Chapter 2 Literature Survey 2. Hence. Therefore. In the near future. the ﬁrst task necessary to study these dynamic eﬀects of the four types of DGs on the distribution lines is to develop models based on the literature and any available operational data on DGs and secondly to utilize that data and the references to develop dynamic models. etc. short construction lead time.1 Introduction The objective of this dissertation is to develop dynamic models of four types of DGs. and low capital expense.

The model is developed to investigate and improve the dynamic control of the gas turbine following a disturbance. The dynamic characteristics of the gas turbine consist of: turbine power Pm in the low pressure turbine. torque THP developed by the high pressure turbine.CHAPTER 2.1 Gas Turbine As gas turbines have fast run-up capacities characteristics. The fundamental equations are: ∆QF = QF − QF 0 ∆NG = µNG ∆X = µX (2. the exhaust gas pressure. Several types of gas turbine models are presented [44][50][107]. Based on the hypothesis that transient thermodynamic and ﬂow processes are considered as quasi-static. A least square cubic spline curve ﬁtting technique is used to ﬁt curves to get the parameters: time constants. µNG and µX are the sensitivity of X an NG to QF . the turbine exhaust temperature Tex and the load torque on the compressor Tload .2) (2. the exhaust gas temperature and exhaust gas power. the dynamic response of gas turbine is described as a series of linearized equations.4) (2. lead-lag ratios and so on. ax is the lead-lag ratio of X. and exhaust gas temperature control. The turbine dynamics are represented by a set of functions which can be determined by steady state measurements. NG is the engine speed. . LITERATURE SURVEY 5 2.3) (2. In [44] a twin shaft gas turbine model is presented. the generator is located in the center with a twin-shaft combustion turbine on each side. In the model. which includes the compressor discharge pressure. TG and TC are the equivalent engine and the combustion chamber time constants.2. This model is suitable for studying the dynamic performance of the twin-shaft gas turbine.5) 1 ∆QF (1 + sTG )(1 + sTC ) (1 + sax TG ) ∆QF (1 + sTG )(1 + sTC ) NG = NG0 + ∆NG X = X0 + ∆X where X is a gas turbine variable vector. The model includes free turbine.1) (2. it is suitable to be connected to weakly supply systems or even used in an isolated operation system. engine speed control. A detailed gas turbine generator model has been presented by Hung [50]. the type of gas turbine model cannot be used in this dissertation research due to the limitation of the measured data. operation curves are not shown in the paper. However. Unfortunately.

simulation and control schemes are developed in the later Chapters.1: General structure of a diesel-engine DG. the diesel engine generator is to supply the emergency power in a nuclear power plant. Besides the function to supply the continuous power to the important customers. 2. Td is the average shaft torque. Sometimes. A diesel engine . drive train. The validation of this model is tested. Another important reason to select this model is that this model is tested in Trinidad and Tobago system [114].1 [18]: In Figure 2. A detailed diesel engine generator including drive train is presented in [18].2 Diesel Engine Diesel engine powered synchronous generator is another type of important DG.1.2. p0 is the zero torque pressure. In [133].CHAPTER 2. The papers [11][18][62] [124] presented the combined multiple renewable energy source systems of the integration of diesel engine with other types of DGs. Ps is the real power to grid. The diesel engine DG model consists of combustion. the simpliﬁed gas turbine model represented in mathematical equations [107] is selected as reference model and is simulated. it is also used for back up. It is suitable to study the dynamic stability of gas turbine in power system. LITERATURE SURVEY grid Diesel Engine P0 6 ~ Synchronous Generator Td Drive Train Ts qd ~ uf AVR Ps uref GOV ωref Figure 2. emergency and stand by. and synchronous generator. the diesel engine DG is working together with the solar electric or wind power system or for a standby alone DG. The mathematical model. Ts is the air gap torque of the synchronous generator. In our research. The general structure of a diesel engine DG is illustrated in Figure 2. qd is the input fuel ﬂow rates.

and τ d is time delay of combustion. The interactions between two types of DGs are assessed. The interaction of multi-type DGs is evaluated under diﬀerent operating conditions.11) where θcl is the torsion angle of the engine-generator shift. The diesel-wind turbine system of [62] supplies power to isolated loads.10) (2. LITERATURE SURVEY is presented by the state equations: qef f = ˙ 1 (−qef f + qd (t − τ d )) τc pe = εKc qef f Tda = Kv (pe − p0 ) 7 (2. However. The impacts of a diesel-wind-photovoltaic generation system is presented in [11]. The impact of the diesel engine and wind turbine DG system on a weak system has been studied. The adaptive fuzzy logic control scheme is implemented to cope with the unpredictable ﬂuctuation of the wind resources. ε is the fuel eﬃciency. Modular subroutines in FORTRAN 77 are called during simulation. In [133]. The drive-train model can be expressed as: ˙ θcl = ω d − ωs ωd = ˙ 1 (εKc Kv qef f − Ccl θcl − (Dd + Dcl )ω d − Kv p0 ) Jd 1 ω s = (Ccl θcl + Dcl ω cl − (Ds + Dcl )ωs − Ts ) ˙ Js (2. ωs is the rotational speed of the electrical generator. The governor and the AVR units .8) where pe is the average engine chamber pressure. ωd is the rotational speed of the diesel engine. Dd and Jd are the damping and inertia of the diesel engine.7) (2.9) (2.6) (2. Ds and Js are the damping and inertia. the model of the diesel engine generator is used to assess the power supplying to the dynamic loads during various transient procedure.CHAPTER 2. Kc is constant relating pressure and fuel consumption. Kv is the stroke volume of the engine. The DGs and load are considered to be located very close to each other. The diesel engine DG is connected with a wind turbine directly to provide reliable power and save fuel. τ c is the time constant of the actuator and combustion process. The software provides a good tool to analyze the dynamic performance and design controllers. qef f is the eﬀective fuel ﬂow rates. The ACSL language based software package has been developed to evaluate the integrated DG system. This model includes the two parts of the general structure: diesel engine and synchronous generator. The distribution system network and the iterations with the substation are not considered.

this model is implemented in Power System Blockset (PSB) [54] as a demo case. The cell’s terminal voltage during a load change is discussed. 2. The operating temperature of PEMFC is about 175◦ F . The SOFC uses a hard ceramic material of solid zirconium oxide as the electrolyte and operating temperatures can reach 1800◦ F .3 Fuel Cell Fuel cells will be important components of distribution system due to their high eﬃciency and low environmental pollution. Due to an electric-chemical process of power generation. and application in PSB. For example. Carbon dioxide emissions of DCFC are much less compared to conventional power generators with a similar power size [77]. low cost and high eﬃciency inverters are required together with acting controllers for fast tracking of real and reactive power demands. However. A transient dynamic model of SOFC was proposed in [42]. LITERATURE SURVEY 8 are also considered. Since PAFC has been commercially used. the model in [133] is considered to be adopted in our DG research. nursing homes. Also. with an improved eﬃciency of about 60%.2. these two types of fuel cells will be studied. and SOFC has the highest potential in large power application. . The electric eﬃciency is more than 40%. utility power plants.CHAPTER 2. solid oxide fuel cell (SOFC) [42][92]. Electrochemical and thermal simulations of a SOFC was reported in [9] to identify the key parameters of fuel cell systems from a single cell to an m ×n cell stack bundle. the net fuel-to-electricity of the direct carbonate fuel cell (DCFC) power plant is about 55-60%. and the operating temperature is 1200◦ F . the PEMFC is very sensitive to fuel impurities. MCFC uses a liquid solution as an electrolyte. Normally. molten carbonate fuel cell (MCFC)[46][77][76][116] and proton exchange membrane fuel cell (PEMFC) [70]. The modular design satisﬁes the versatile demands of customers. the generation electric eﬃciency is about 60%. Fuel cells must be connected to the distribution system via DC/AC inverters. The SOFC can be used in large-power applications such as central electricity generation station. Due to the veriﬁcation of the ﬁeld measurement. The transient process of the drive train is ignored. Several types of fuel cells have been reported in the literatures: phosphor acid fuel cell (PAFC) [106][135]. The results of the ﬁeld measurement verify the model. Therefore. The PAFC has been commercially used in hospitals. there is no noise develop usually in mechanical members of conventional generator. All of these features will without any doubt lead to their wide application in the power industry in the near future. etc.

2.2: Structure of a fuel cell power plant.3. a wind turbine is connected to an induction generator to implement the energy conversion. In [1].2). Another wind turbine model is presented in [62]. torque computing unit. The inverter serves as the interface between the fuel cell and the power distribution system. Wind turbines can either be connected to a synchronous generator [68][88] or an induction generator [1]. The wind turbine system of [88] can be summarized as shown in Figure 2. The fuel cell control is achieved by adjusting the input volume of gas and air and controlling real power output.CHAPTER 2. A simpler wind turbine induction generator model is developed in [59] as it was speciﬁcally used for dynamic studies of dispersed wind turbine applications.12) . 2.4 Wind Turbine Wind turbines are widely used as they are environmental friendly energy sources. The model includes the synchronous generator. In these papers. The model is: λ= ΩR Vω (2. torsional system and the control loop using output-feedback design. a controllable horizontal wind turbine model is adopted. It is controlled in order to provide real and reactive power set point tracking and to adjust the power factor. Fuel Cell Plant Balance of Plant Fuel Cell Stack Power conditioning unit AC Power Power Grid P and Q Plant Controller Commands Network Interface Controller Network Parameters Figure 2. A static var compensator (SVC) is used to regulate its voltage. This model is suitable for analyzing dynamic characteristics of wind turbines and designing control. LITERATURE SURVEY 9 Reference [92] proposed a dynamic model for a SOFC stack and gave the general structure of a fuel cell power plant (see Figure 2.

CHAPTER 2. an algorithm is given to provide the two wind speed components separately. ¸ π(λ − 3) Cp = (0.3β Tm = 1 2 2 ρARCp Vω · (2. Tω is the wind torque. MT . DG and DT are the generator and turbine inertias and damping coeﬃcients. LITERATURE SURVEY XT Wind and Blode Torque Calcs u=(s.16) δ = −ωG + ω T K DT DT 1 (2. The simpliﬁed wind turbine induction generator model [59] is expressed as: ωG = ˙ 1 −(DG − DT ) (DG − DT ) 1 ωG + ωT + TG − PG MG MG MG MG (2. u is the turbulent component of the wind. In these papers. [18].0167β) sin − 0. and Tm is output mechanical torque of the wind turbine. ∆Vt. Vfd)T Out put Feedback Control 10 2-mass torsional system Synchronous Generator Y=(∆ω.3: A wind-synchronous generator with a control loop.13) (2. Ω is the mechanical angular velocity. . R is the wind turbine rotor radius. In this research.17) ωT = ˙ ωG − ωT + Tω − MT MT MT MT where: MG . To implement the multiple renewable energy source system. K is the torsional spring constant. Cp is the coeﬃcient of nondimentional curves of the power. β is the blade tip pitch angle.14) λ where: λ is the blade tip speed ratio.15) ˙ (2. integration of multiple types of DGs are discussed in [11]. the wind turbine is considered to work at the rated power regime and the disturbance of the wind model is beyond consideration. ρ is the air density. A wind model is expressed by the instantaneous wind velocity [62]: Uw = U + u(m/s) (2. the coordination of the wind turbine and the diesel engine is investigated. [62] and [124]. A is the set area by the blades.00184(λ − 3)β 15 − 0.18) where U is the mean wind speed. Vω is the wind velocity. ∆Pe)T Xe Grid Figure 2. In [62].44 − 0.

CHAPTER 2. LITERATURE SURVEY

11

2.3

Control Strategies for DGs

To improve the dynamic stability of the distribution system with multiple types DGs, control strategies for DGs are investigated. As a relatively new topic, most of the literatures found concentrates on the control theory application on the transmission level, for example, [5], [22], [26], [30], [71], [126] and [131]. However, power distribution systems have their own unique features: large capacity substation, relatively small coverage area and unbalanced loads. These concerns have to be addressed. Both decentralized control and global control schemes are investigated. Since the substations of distribution systems have a relatively large capacity, it is usually considered as an inﬁnite bus. The dynamic stability is the major stability problem rather than the transient stability. At present, the residue and the damping torque analysis methods are mature and have been applied to design dynamic stability controllers as PSS and FACTS devices. The centralized control can eﬀectively improve the dynamic stability of the whole system. The signal communication between remote areas is a constraint on developing controllers. However the power distribution system is located in a relatively small area, hence the signal communication is practically feasible. The centralized control scheme will be considered as the major type of control to improve oscillation damping of distribution systems with DGs.

2.3.1

Linear Analysis

The purpose of the DG controllers is to improve the dynamic stability of the system. The linear analysis methodology has long been used to study the stability problem by eigenvalue calculation and critical oscillation modes identiﬁcation. The authors of [32] and [115] proposed indices to rank eigenvalues. The eigenvalue with the largest index is the most critical one. By using this approach, the most critical oscillation modes can be identiﬁed during control design.

2.3.2

Model Reduction

Model reduction is considered as the ﬁrst step in modeling large-scale systems and designing feedback controllers. Applications can be viewed in [32], [23] and [53]. Feliachi proposed an algorithm based on optimizing output performance index in [32]. The outputs of the system are the state variables. The method has been implemented in [53]. By reducing the model to a simpliﬁed reduced order system model, eigenvalue assignment method is applied to design a controller and promising results have been achieved.

CHAPTER 2. LITERATURE SURVEY

12

Another eigenvalue assignment based on a matrix reduction technique with application to power systems was proposed by Choudhry [23]. The power distribution system has the distinct feature of strong connections between components. Hence, a subgroup of eigenvalues that can represent the system performance with little errors cannot be found since all components may have similar contribution to eigenvalues. Based on these references, the author obtained a reduced system and designed a controller using eigenvalue assignment method. This controller can move certain eigenvalues that have been included in the reduced model perfectly to destined locations. However, when the controller is applied in the original system model, some eigenvalues not included in the reduced model shifted to the right-half plane which worsen the dynamic performance of the system. In addition, the distribution test system is of order 50, a size that can be easily handled. Therefore the full system model is used to design controllers.

2.3.3

Control Schemes

Residue based indices were proposed in papers of Martin [78] and Yang [131]. This method is suitable for designing single input single output (SISO) controllers at a speciﬁc operating point. The centralized control can eﬀectively improve the dynamic stability of the whole system. For a large-scale system, signal communication is the constraint to develop such controllers. However, the power distribution system is relatively small and signal communication can be easily achieved. The centralized control scheme will be considered as the major type of control to improve oscillation damping of the distribution system with DGs. Output feedback control is one kind of centralized control scheme using information from diﬀerent locations as input. This scheme will be adopted in this dissertation to design controller for wind turbine controller. Both decentralized and centralized controllers as mentioned above are designed for a speciﬁc operating condition and may not function well when the operating condition changes. In reality, power transactions between areas and loads vary considerably and the operating condition changes frequently. Therefore, uncertainty in operating conditions has to be addressed in the control design process. A structured singular value approach was presented in [26] to address this issue by designing a robust controller which is able to accommodate the uncertainties. In this dissertation, uncertainty will be modeled and the capability of the controller to handle uncertainties will be discussed to study its robustness.

CHAPTER 2. LITERATURE SURVEY

13

2.3.4

Coordinated Control

The issues of coordinated control have been addressed in transmission systems [71][34] where PSS and FACTS devices such as TCSC and SVC have been installed. The design of one controller to improve the damping of a certain oscillation mode should not deteriorate the damping of other modes. A widely used coordinated control strategy is the multivariable optimal control. With diﬀerent DGs in the system, diﬀerent kinds of controllers will be installed. The coordination among controllers is very important for the system to function properly and have a good dynamic performance.

2.3.5

Robust Analysis of the Load Parameter Uncertainty

The robustness of designed controllers will be veriﬁed by testing loads with parameters uncertainties and/or diﬀerent operating conditions. The inﬂuence of the load parameters on the dynamic stability of the system has been studied in [85]. An additional SVC stabilizer (ACS) has been tuned for robust control. However, in [85], only voltage dependant static loads are involved. The eﬀect of load models on stability of a modulated AC/DC system is investigated in [29]. An eigenvalue sensitivity calculation method is developed to predict eﬀects of load characteristics on system stability. A method to design robust modulation controller over a wide range of load characteristics for AC/DC system is proposed in [22]. In our research, the parameter uncertainties of dynamic loads are taken into consideration during the controller design.

2.4

Power Electronics Application

Power electronics are commonly used in power transmission system as they are the main components of Flexible AC Transmission Systems (FACTS). Power electronics are also applied in existing power distribution systems to improve power quality by applying power conditioning units [48]. Unlike the customer power, in our DG project, power electronics are applied as the important part of the DGs as DC/AC the interface of the fuel cell, or as the var compensator for the wind turbine. The rapid response feature of the power electronics enables them as the good real/reactive power control devices.

2 Current/Voltage Source Inverter (CSI/VSI) The requirements of the interface between the fuel cell system and the power distribution system include low cost. is realized through current feedback. the voltage regulator is of the integral.4. 4) Facilitate synchronizing a turbine-generator. diﬀerent ﬁring angles is illustrated.3 Thyristor Controlled Braking Resistor (TCBR) A TCBR is a simple. the inertia modes are damped eﬀectively. Two basic modeling approaches with diﬀerent methods to realize the slope function are recommended. 3) Damp subsynchronous resonance (SSR). 2.a gain of 100 pu Bsvc /pu ∆V on the SVC base means a 1% slope. the SVC is widely used in power systems to supply reactive power. the functions of TCBR are summarized as: 1) Damp low frequency interconnection oscillation. the voltage regulator is of the proportional type and the gain. or proportional plus integral type and the slope. adjustable power factor and low line-current distortion [89]. A practical model for advance power ﬂow is presented in [93]. This kind of inverter has been used to connect photovoltaic arrays to the AC grid [6]. . Diﬀerent approachs/realization can be found in the literatures. In [128].4. high reliability and eﬃciency. In [97]. an average inverter model is described to represent the static synchronous compensator (StatCom) which converts the DC battery energy storage to the AC transmission system.1 Static VAR Compensator (SVC) As a mature technology. KR . The “inverter ﬂux vectors” with additional reactance are used in [6] to control real and reactive output power. is the inverse of the slope . It is simpler and easier to implement. In [47]. KSL . A vector space controlled inverter based on GTO switches fulﬁlls the above requirements [6][17]. 2) Prevent ﬁrst cycle swing transient instability. A closed-loop quasi-optimal switching schema was reported in [97]. a variable structure controller (VSC) was implemented to damp the interconnection oscillation in a multi-machine system. With the VSC controller. In [132]. a control cost index was used to optimize the control strategy. etc.CHAPTER 2. LITERATURE SURVEY 14 2. highly reliable. The controlled susceptance vs. low cost and very eﬀective Flexible AC Transmission Systems (FACTS) device. In method one.4. A completely current regulated PWM_VSI (pulse width modulation voltage source inverter) is described in [87]. Many control strategies have been proposed. A modular structure for SVCs for the dynamic study is proposed in [56]. 2. In method two.

.CHAPTER 2. LITERATURE SURVEY 15 This TCBR and VSC control strategy will be implemented during the three-phase system study to improve oscillation damping of the whole system [81].

The total load and voltage of this system are 20. loads.2. Some of the results are illustrated in this chapter. the power system toolbox (PST) is utilized as the tool for study and analysis. and others will be represented in the following chapters together with the continued research. .0kV respectively [65][66].1. the actual suburban system I (SI) is represented in this chapter to brieﬂy demonstrate the utilities related DGs research.1 Introduction In the distributed generations (DGs) research. A series of case studies is conducted to evaluate the models.5MVA and 23. As an example. The substation is modeled as a combination of a synchronous generator and a transformer.2 3. and DGs of real system are discussed. lines. The DGs are modeled as synchronous generators with simpliﬁed turbine and excitation.1 Models of An Actual Distribution System Suburban System I (SI) The one-line diagram for the SI distribution system is shown in Figure 3. The impacts of DG on line voltage and system stability are evaluated including the eﬀects of several groups of factors.16 Chapter 3 Preliminary Research of DGs 3. two years have been invested in a DG project evaluating various DG sources connected to computer simulation models of actual distribution systems. The SI system is a typical suburban system. In this preliminary DG research. 3. The simulations of the substation.

1) Substation Traditionally. The bases were used to model all the remaining details of all the lines and transformers in order to derive the new single line equivalent. To investigate dynamic characteristics of the distribution system with DG. 2) Loads The loads along the distribution lines were represented as static loads with a ﬁxed power factor of 0.85 lagging.05 (base on DG2) 5 ~ 501 516 515 430 427 401 519 11 17 4 Rt+Xt=0. PRELIMINARY RESEARCH OF DGS 12 6 Rt+Xt=0. The form of the substation model used here consists of a synchronous generator and a transformer. 3) Distribution Line . The distribution line is represented by the equivalent impedance which reﬂect the active and reactive losses. Sbase = 20.05 (base on DG1) 3 ~ DG2 608 601 DG1 603 617 715 104 801 103 Xt=0.5M V A as the system base. 3. a model for a substation for use with a distribution system is a voltage source in series with an impedance. The load of the single line equivalent system was based on the sum of loads in the unreduced (e. Therefore.2). Using Vbase = 23kV .g.1 (base on 42.0MVA Figure 3.05+j0. The synchronous generator is assumed to have a very large inertia and capacity. the single line equivalent represents all the losses.05+j0.2 Models The SI distribution system is simulated in PST. a new distribution substation model is developed [80] and the model will be described in detail in Chapter 4.2. three-phase) system (see Figure 3.CHAPTER 3.1: One-line diagram of the SI distribution system with DGs.0MVA) Substation ~ 2 1 42. all the other lines and loads are represented in per unit on this base.

**CHAPTER 3. PRELIMINARY RESEARCH OF DGS
**

Z4 V,δ Z ~ to other line Z1 Z2 Z3 Load 4

18

Equivalent Source

Load 1

Load 2

Load 3

Combine Multi-Lines and Mult-Loads to Equivalent Single-Line, Single-Load Circuit

V,δ Z ~ Zequ ∑ Load Equivalent Source to other line

Figure 3.2: Diagram illustrating the procedure used to combine multiple lines and multiple loads into an equivalent single line and single load equivalent circuit. The distribution line considered is a three-phase balanced line and is modeled as a singlephase line with static parameters. The original three-phase line was represented by the single-phase line as shown in Figure 3.3. 4) Distributed Generation (DG) Each DG is modeled as a synchronous generator with simpliﬁed turbine and excitation, and a transformer. A transformer compatible to the DG capacity connects the DG to the distribution line. The simpliﬁed turbine governor model and exciter model are used for the DGs [95].

3.3

Test Cases and Simulation Results

A set of test cases was designed to investigate the transient stability and dynamic performance of the system based on diﬀerent load characteristics. In Figure 3.1 two DGs (synchronous generators) are located at buses 11 and 12. Fault locations for the diﬀerent experiments are next to the DG, such as at buses 11 and 12, and in the middle of the distribution feeder line, such as at buses 103 and 515. Referring to the protection manual [25], a fault clearing time of 0.084s, and a reclosing time of 0.64s were chosen. Several simulations with other fault clearing and reclosing times are done and all the results are also compared. The diﬀerent impacts of the DG on the distribution system are classiﬁed as: · Diﬀerent fault location;

CHAPTER 3. PRELIMINARY RESEARCH OF DGS

19

to other line V,δ Z ~ Phase B Equivalent Source Phase C ZC ZB Phase A ZA Load 1 to other line Load 2 to other line Load 3

Original Three-Phase to Single Phase Equivalent Circuit

V,δ Z ~ Zequ ∑ Load Equivalent Source to other line

Figure 3.3: Diagram illustrating the procedure for representing three-phase as a single-phase equivalent circuit. · · · · Diﬀerent fault clearing time; Diﬀerent reclosing time; Diﬀerent power size of DGs; Diﬀerent DG location.

In the following ﬁve subsections, the simulation results illustrate the impacts of the part of factors. Plots of voltage proﬁle, frequency responses, rotor angle, and real power are used to illustrate the transient process.

3.3.1

Fault Location

Worst cases for diﬀerent fault locations appear to be those in which the fault is near the DG source as would be anticipated. Cases 9, 22 and 30 shown in Figure 3.4 and a series of other simulations tabulated in Table 3.1 displayed this result.

3.3.2

Fault Clearing Time

Generally, it would be anticipated that shorter fault clearing times would contribute to reduced power system disturbances. Figure 3.5 shows the voltage proﬁle of the substation with diﬀerent fault clearing times. During the fault time, the voltage of the substation falls below 0.8pu. Such

CHAPTER 3. PRELIMINARY RESEARCH OF DGS

20

Table 3.1: The tests G1, G2 P owerSize Case (M W ) 1 0.8, 0.8 2 1.6, 0.8 3 0.2, 0.2 4 0.1, 0.1 5 0.4, 0.4 6 0.4, 0.2 7 0.1, 0.1 8 1.6, 1.6 9 2.4, 2.4 10 2.4, 2.4 11 2.4, 2.4 14 1.6, 1.6 15 0.8, 0.8 16 0.4, 0.4 17 4.0, 4.0 18 6.0, 4.0 19 6.0, 6.0 20 2.4, 2.4 21 2.4, 2.4 22 2.4, 2.4 23 2.4, 2.4 24 2.4, 2.4 25* 2.4, 2.4 26* 2.4, 2.4 27* 2.4, 2.4 28* 2.4, 2.4 29* 2.4, 2.4 30 2.4, 2.4 31 2.4, 2.4

on the SI distribution system. Clear Reclose F ault T ime time Location (sec) (sec) 11-427 0.084 0.64 11-427 0.084 0.64 11-427 0.084 0.64 11-427 0.084 0.64 11-427 0.084 0.64 11-427 0.084 0.64 11-427 0.084 0.64 103-104 0.084 0.64 103-104 0.084 0.64 103-104 0.15 0.64 103-104 0.30 0.64 519-401 0.084 0.64 12-430 0.084 0.64 12-430 0.084 0.64 103-104 0.084 0.64 103-104 0.084 0.64 103-104 0.084 0.64 515-516 0.30 0.64 515-516 0.15 0.64 515-516 0.084 0.64 515-516 0.084 1.28 515-516 0.084 2.56 515-516 0.084 0.64 515-516 0.15 0.64 515-516 0.30 0.64 515-516 0.084 1.28 515-516 0.084 2.56 11-427 0.084 0.64 11-427 0.084 2.56

Figure 3.CHAPTER 3. fault clearing times. the worse the voltage quality is.64s and 2.3 Reclosing Time The impact of the reclosing time is investigated in cases 30 and 31.95 c ase 9 59. respectively. a low voltage may have undesirable impacts on other loads connected to the substation (e. These simulation results show: . fault locations.4: Frequency of DG1 with diﬀerent fault location (case 9. 3. case 22.75 c ase 22 c ase 30 frequency (Hz) 0 2 4 6 8 10 12 time (s) Figure 3. 3. The longer the fault lasts.05 60 59.3. case 30).15 60. Therefore. PRELIMINARY RESEARCH OF DGS 21 60.85 59.1 60.25 60. and 31 are 0. Reclosing times for cases 30. it may cause them to be disconnected). DG power sizes.g.9 59. All other simulation conditions are the same. reclosing times. faster fault clearing times will improve the power quality. Five selected groups of cases are presented here to illustrate the impacts of these factors. variations in all of which were evaluated via 31 diﬀerent simulation cases (see Table 1).56s. and DG locations.6 shows that for larger reclosing times the amplitude of the resulting damped oscillation is larger and may even result in divergence (see case 31).8 59.2 60.4 Summary Whether a distribution system remains stable or not when DG is added depends on contingencies such as faults types.

6: Frequency of DG1 with diﬀerent DG1’s location (case 22.8 0. case 25). PRELIMINARY RESEARCH OF DGS 22 1.85 0.8 0 2 4 6 8 10 12 time (s) Figure 3.95 59.75 0.95 0. 60.1 c as e 25 60.85 59.CHAPTER 3.05 case 22 1 case 20 0.9 voltage ( pu ) 0.65 0 2 4 time (s) 6 8 10 12 Figure 3.9 59.5: The voltage magnitude of the substation with diﬀerent fault location (case 20.05 frequency (Hz) c as e 22 60 59.15 60. .2 60. case 22).7 0.

CHAPTER 3. The most important factors aﬀecting the degree of system relative stability are the fault location. The DG location is an important factor in determining frequency deviation and the system. thus the relative stability characteristics of the system. frequency recovery time. reclosing time and the DG locations. The fault clearing time and the DG size also aﬀect the stability of the SI distribution system. PRELIMINARY RESEARCH OF DGS · · · · 23 Fault location is an important factor aﬀecting the relative stability of the distribution Fault clearing time impacts the most on voltage quality along the distribution system. . Worst case faults are those occurring near the DG sources. but to lesser degree than the above three factors. Shorter fault clearing times improve the voltage quality. Shorter reclosing times improve system stability.

In this chapter. simulate these models in . the synchronous generator was represented by the simpliﬁed transient/subtransient model. such as gas turbine. This matrix/vector based method can avoid the numerical algebraic loop problems. In this chapter. At the same time.1 Introduction In the preliminary research. As a new solution. In the DGs research.24 Chapter 4 DGs Simulation and Linear Analysis with Simulink 4. the following tasks have been done: 1. two of the four types DGs model: gas turbine and diesel engine DGs are developed. The program structure of this simulation package is presented in Appendix C. a software package is developed in Matlab/Simulink step by step to study the distribution system with DGs. the SI system and the distributed generations (DGs) are represented in the power system toolbox (PST). Develop the gas turbine model and diesel engine model. The full-order synchronous model is chosen to be connected with gas turbine and diesel engine. the PST is diﬃcult to insert some speciﬁc electric components. As discussed before. the synchronous machine model is simulated using the matrix/vector concepts in Simulink. In the early power system studies. voltage source inverter. In the following chapters. Matlab/Simulink provides a tool to obtain the linearized model from the nonlinear simulation models in Simulink. it has become apparent that these simpliﬁed models are not adequate to describe the dynamic response of a complicated power distribution system.

The 6th -order state equations of the electrical part can be expressed as: r vqs = −rs ir + qs r vds = −rs ir − ds ωr r p ψ + ψr ω b ds ωb qs (4.8) . 4. The 2nd -order state equations of the mechanical part are: 1 p (Te − Tm ) ωr = − ωb 2H pδ r = ω r − 1 (4. Analyze the eigenvalues of the distribution system with DGs and each of its components. Simplify the IEEE-13 node feeder. These two types of DG are considered as the basic DGs and their models will be included in the test system.2 System Models To study the dynamic characteristics of the gas turbine and the diesel engine DGs. 4. 25 2. Form a way to compute the ﬁrst-order eigenvalue sensitivity via Simulink.1) (4.1 Full-Order Synchronous Generator The full-order synchronous generator model consists of 6th -order electrical equations and 2nd -order mechanical equations.2. then connect them to the gas turbine. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK Simulink. 5. the full order synchronous generator is developed in Simulink. while the substation is modeled as a simpliﬁed two-axis synchronous generator. Represent the substation by a simpliﬁed two-axis synchronous generator model.5) (4.CHAPTER 4.2) (4.7) (4. 3.3) (4. and simulate it in Simulink. diesel engine and excitation system to represent the DGs. Investigate and simulate the full-order synchronous generator in Simulink. 6. 4.6) ωr r p ψ qs + ψ r ωb ωb ds p 0r 0 vf d = rf d i0rd + ψ 0rd f ωb f p 0r 0 vkd = rkd i0r + ψ 0r kd ω b kd p 0r 0 vkq1 = rkq1 i0r + ψ 0r kq1 ωb kq1 p 0r 0 vkq2 = rkq2 i0r + ψ 0r kq2 ωb kq2 where p = d dt .4) (4.

10) 0 pEq = 1 0 (−Eq + EF D + (xd − x0 )Id ) d t0 d0 (4.2 Substation Model To investigate dynamic characteristics of the distribution system with DG.CHAPTER 4.2. the terminal voltage magnitude of the synchronous generator is considered to be constant. where ωb corresponds to rated or base frequency. the capacity of the substation can be so adjusted. The ωb is used to calculate the inductive reactances.1). The form of the substation model used here consists of a synchronous generator and a transformer (see Figure 4. The synchronous generator is assumed to have a very large inertia and capacity. its frequency is assumed to be constant requiring that the inertia of the synchronous generator must be suﬃciently large.11) . Second. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK 26 Substation Bus Transformer ~ Synchronous Generator Figure 4. There are three reasons for this assumption: ﬁrst. similar to actual practice. a new distribution substation model is proposed in [80]. Third.1: Diagram of the substation model of the distribution system.9) The detailed full-order synchronous generator model is described in Appendix A. 4. since the substation can provide adequate power for the distribution system. with small transient reactance and large capacity. The state equations of the electrical part are: 0 pEd = 1 0 (−Ed + (xq − x0 )Iq q t0 q0 (4. and the torque in per unit is: Te = ψ r ir − ψ r ir qs ds ds qs (4. since the substation is connected with the grid.

14) The transformer is modeled as a pure reactance with capacity equal to the capacity of the 4.3(Wf − 0. The input signal is synchronous generator speed N . valve position. limits.23) + where N : is the synchronous generator speed in per unit. Wf : is the fuel ﬂow. is 43MW [63].CHAPTER 4. This gas turbine model is suitable to investigate the dynamic stability of gas turbine generator in power distribution system. The turbine output is: Pmech = f (N. The capacity of the synchronous generator is determined by thermal constraint of the 138kV grid connected to the SI distribution system.0M V A (4.15) (4. A simpliﬁed dynamic single-shaft gas turbine model is shown in Figure 4. the rating of the 138kV is three times the surge impedance loading [31]: M V A = 3 ∗ 43.0 = 129. and output is control signal Pmech to a synchronous generator. IC2: are initial values of valve position and fuel pump.3 Gas Turbine A gas turbine driven synchronous generator is one of the important DGs studied. The power carried by the 138kV.2 [114] [107].16) . DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK The electrical torque output is: 0 0 Te = Ed Id + Eq Iq − (L0 − L0 )Id Iq q d 27 (4. Then.12) The state equations of the mechanical part are: pω = 1 (Tm − Dω − Te ) tj pδ = ω − 1 substation. The gas turbine model consists of speed governor.2. Wf ) = 1.13) (4. 1−N 2 (4. IC1. and turbine. fuel pump. called surge impedance loading or natural load.

23 + + N X 0. and the output is the control signal Pmech to a synchronous generator.2. the following assumptions are taken into consideration to transfer the three-phase unbalanced distribution system .3 [133].4 Diesel Engine A diesel engine driven synchronous generator is the second type of DG to be studied. 4. The input signal is the synchronous generator speed ω in per unit.5 Average Model of IEEE-13 Node Feeder To simulate the IEEE-13 node feeder and use the linearization tool of Simulink. ω_ref + Control System -(1+T3s) 1+T 1s+T2T1S2 Actuator K(1+T4s) (1+T5s)(1+T6S) Limits Integrator 1 s Engine e-sTD Pmech X ω IC Figure 4.2: Diagram of a simpliﬁed gas turbine model. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK N_ref + Governor Kd 1+T gs 28 P-Limit 0.77 N + IC1 1 s Tv IC2 Valve Position Fuel Pump + 1 s Wf f(N.CHAPTER 4.Wf) Turbine Pmech Tf Figure 4. The diesel engine model is shown in Figure 4.3: Diagram of a diesel engine and a governor model. The IC is the initial value of the integrator.2. 4.

4: IEEE-13 node distribution system with a gas turbine DG and a diesel engine DG.3 Building Synchronous Generator Model using Matrix Concept in Simulink Matrix and vector are basic concepts used in Matlab/Simulink. two or single phase loads of each bus are represented as single load. Transformer and voltage regulator are represented single-phase reactance. The substation and the DGs are represented in synchronous p-q reference frame. 4. Vector computation is one of the features of Matlab.CHAPTER 4. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK 29 Subst at ion 650 646 645 632 633 634 Diesel Engine Syn Gen 611 684 671 692 675 680 652 Gas turbine Syn Gen Figure 4. 4. 2. each diﬀerential equation can be represented by a transfer function composed of an integral unit. With a set of diﬀerential equations. For instance. While an input of one integral unit might be the function of the outputs of other . When building models in Matlab/Simulink. 3. Using matrix/vector concepts not only simpliﬁes problems but also contributes to time saving in debugging. the load of bus 675 is 485+j190+68+j60+290+j212 3 = 281 + j154. Three. Dynamic machinery models are usually expressed in a set of diﬀerential equations. This concept is to be used in simulation in full-order synchronous machine. into a balanced single-phase system: 1. The one-phase impedance of each feeder is used to represent the feeder. the value is the average over three. a set of integral units will be used.

The equations of a full-order synchronous machine model in the arbitrary reference frame can be written in term of the currents as shown in equation (4.1-4. where x = ˙ [iqs .CHAPTER 4. Hence one integral unit is enough to express this model. Using matrix concepts. ids . a set of diﬀerential equations will be expressed as: x = Ax + Bu ˙ (4. This equation can be further written into: x = Ax + Bu. i0 ]T kq1 kq2 f kd . i0 d . The equations will be presented blow in terms of equation (4. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK 30 integral units. the Simulink model can be very complex and unorganized with lines interweaving. p −rs − ω Xq where: b b ωr p Xq −rs − ω Xd ωb b b b b b p p p 0 0 0 rkq1 + ω Xkq1 Xmq 0 0 − ω Xmq ωb b b M = p p p 0 0 0 rkq2 + ω Xkq2 0 0 − ω Xmq ωb Xmq b b p p p Xmd 0 0 0 − X0md ( ω Xmd ) 0 0 (rf d + ω Xf d ) X0md ( ω Xmd ) 0 rf d rf d rf d b b b p p p 0 0 0 0 Xmd rkd + ω Xkd 0 − ω Xmd ωb b b Assume that the reference frame is a synchronous reference frame and that all quantities vds 0 vkq1 =M 0 vkq2 0 vf d 0 vkd vqs · ids i0 kq1 i0 kq2 i0 d f 0 ikd p ωb Xmq − ωr Xmq ω ωr ωb Xmd p ω Xmd iqs (4. The dynamic model of synchronous machine (see equation 4. i0 .18).6) can be equivalently expressed as the current state variables equations[69]. The Simulink expressions are also shown blow which will use voltages as inputs and currents as the outputs of the model block.18) − ωr Xd ω p ωb Xmq − ωr Xmq ω ωr ω b Xmd p ω Xmd are in per unit value. A and B are matrices. i0 .17).17) where x and u are vectors.

1 − ω Xq 0 1 ωb Xmq 1 ωb Xmq 0 0 0 Xmd 1 ( ω Xmd ) 0 rf d b 1 0 Xkd ω b 1 ωb Xmd 0 In this model block.5: Using matrix/vector to represent the synchronous generator model in Simulink. To analyze the dynamic characteristics and design the . −1 b 1 1 0 0 0 − ω Xd ω b Xmd b 1 1 1 0 0 0 − ω Xmq ω b Xkq1 ωb Xmq b B= 1 1 1 0 0 0 − ω Xmq ωb Xmq ω b Xkq2 b Xmd 1 1 0 0 ( X0 ) 0 − X0md ( ω Xmd ) 0 rf d rf d ω b f d b 1 1 0 0 0 − ω Xmd ω b Xmd b ωr ωr −rs − ωr Xd 0 0 ωb ωb Xmd ω b Xmd ωr X 0 0 −rs − ωr Xmq − ωr Xmq ω q ωb ωb b 0 0 rkq1 0 0 0 0 A = −B · 0 0 0 0 rkq2 0 0 0 0 0 0 0 Xmd 0 0 0 0 0 0 rkd The Simulink model for (4. inputs are voltage and rotor speed.5. The induction machine model simulated with the matrix/vector structure is presented in Chapter 8. 4. It saves not only model building time but also debug time. the DGs can be treated as linear models for controller design to improve the damping of the distribution system. output is current vector. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK 31 Figure 4.CHAPTER 4.18) is very simple shown in Figure 4. This model is quite simple and easy to understand.4 Models Linearization Under the operating point. This model can avoid the numerical algebra loop.

19) is the rotor of the synchronous generator. D = 0. the system plant (A) consists of: the synchronous generator (A11). the linearized models of each components are obtained via the Simulink. Xlf d = 0. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK 32 ∆Iq ∆Id ∆Efd Substation ∆ω ∆E'q ∆E'd 1 s ∆δ Figure 4.000929. dynamic controller of the distribution system with DGs.8958.1.1198. 4.6.1). Xls = 0.01334. Tq0 = 0.0. the distribution system consists of the gas turbine synchronous generator (DG1).CHAPTER 4.2 Linearized Gas Turbine Synchronous Generator The linearized gas turbine powered synchronous generator model (DG1) is illustrated in Figure 4. Td0 = 8. xq = 0.8. valve position (xval ) and fuel pump (xf p ).19. 0 Synchronous generators: rs = 0.003. PDG2 = 0. Substation: H = 2364. rkq1 = 0.4. Sb = 4. r = 0.0.8. A21). rkd = 0.8125.08125. Xq = 1. These parameters and the parameters of gas turbine and diesel engine are summarized in the appendix. The substation is considered as a reference bus and the two DGs are considered as PV bus.8645.0M V A.1 Linearized Substation Model The linearized substation model can be represented as the diagram in Figure 4.00178. the gas turbine (A22) and the interactions between the gas turbine and the synchronous generator (A12. The linearization model is obtained from nonlinear model in Simulink directly.4. Vb = 4. In this chapter.7. xd = 0. 0 0 0 0 0 Xlkq1 = 0. rf d = 0. VDG1 = 1. (see Table B.19). Xd = 1. PDG1 = 0.1. Xlkd = 0.96.6: Diagram of the linearized substation model of the distribution system.1414. VDG2 = 1. In the linearized model (4.16kV . x0 = x0 = q d 0. .1578. The three state variables of the gas turbine are: governor power (xgov ). 4. The reference of equation (4.0875. diesel engine synchronous generator (DG2) and substation.

the output of the integrator (xint ) depends on the four states. xcon2 ).19) 4. where the eigenvalues of the gas turbine synchronous generator under the speciﬁc operating point is: ∆ω ˙ ∆ir ˙ d ∆ir ˙ fd ∆ir ˙ kd ˙q ∆ir ˙ ∆ir kq ˙ ∆δ . ∆xgov ∆xval ∆xf p ∆ω r ∆vq r + B · ∆vd r ∆vf d (4. .8. The diesel engine unit has four independent state variables: two controller states (xcon1 ... ∆xgov ˙ ∆xval ˙ ∆xf p ˙ ∆ir d ∆ir fd ∆ir kd ∆ir q A11 A12 r · ∆ikq = A21 A22 ∆δ .CHAPTER 4.7: Linearized gas-turbine powered synchronous generator. actuator state #1 (xT F 1 ) and actuator state #2 (xT F 2 )..3 Linearized Diesel Engine Synchronous Generator The linearized diesel engine powered synchronous generator model (DG2) is shown in Figure 4.. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK ∆veq ∆ved e->r ∆vrq ∆vrd ∆vrfd ∆δ ∆Tm Synchronous generator ∆ird ∆irfd ∆irkd ∆irq ∆irkq ∆xgov ∆xval ∆xfp ∆δ ∆ied r->e 33 ∆ω ∆ieq ∆ω Gas Turbine Figure 4. Under speciﬁc operating point..4..

983 λ3 λ4.1222 λ9 -2.7 -0.6572i -0.73761i λ6.69465 ± 2.4543 λ10 ∆veq ∆ved e->r ∆vrq ∆vrd ∆vrfd ∆δ ∆Tm Synchronous generator ∆ω ∆ird r->e ∆irfd ∆ied ∆irkd ∆irq ∆irkq ∆ieq ∆ω ∆xcon1 Diesel Engine ∆xcon2 ∆xT F1 ∆xT F2 ∆xint ∆δ Figure 4.2 -3.5 -20.1: Eigenvalues of the linearized synchronous generator with gas-turbine.CHAPTER 4. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK 34 Table 4.8: Linearized diesel engine powered synchronous generator. Eigenvalues λ1.2937 ± 377. NO.06i -30.028 ± 0.37188 λ8 -1. .

5.3: Eigenvalues of the linearized diesel engine DG.2.1 Eigenvalue Analysis Eigenvalue of the Distribution System with DGs The eigenvalues of each components are discussed in above section.CHAPTER 4.2: Eigenvalues of the linearized synchronous generator connected to diesel engine. Eigenvalues 0 λ1 —26.4116 λ6 -1. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK 35 Table 4.042 λ2 -111. The results show the numbers of each component as: .5 -0.065 λ7 Table 4. ∆x ˙ con2 ∆xT F 1 ˙ ∆xT F 2 ˙ ∆xint ˙ ∆x con2 = Adies · ∆xT F 1 ∆xT F 2 ∆xint The zero eigenvalue is corresponding to the ∆xint . ∆xcon1 ˙ ∆xcon1 (4. NO.2935 ± 377. Eigenvalues λ1.2 -3.5 4.70467 ± 2. The eigenvalues of the synchronous generator of DG2 are given in Table 4.144i The linearized synchronous generator model is diﬀerent from the model of DG1 since they have diﬀerent operating points. The linearized model of the diesel engine is shown in equation (4.5 -25 ± 66. It means the integrator is only an 4.20) where the eigenvalues of the diesel engine are given in Table 4.20).3.11 λ3 λ4. NO. output of the isolated linearized diesel engine model.983 λ3 λ4.4787i -0.06i -30.

024 ± 0.913 ± 64.225 λ21 -2.387i λ6.48735 λ26 -0.343 λ10 -25.081228i -0.41639 λ25 -0.8898i λ19.18 -0.2691 ± 10.0020607 ± 0.487i λ14.968 λ11 -20. Table 4.8487 ± 3.4846i λ17. NO.20 -4.4 -108.54i λ1.1171 ± 8.66±928.606±440. the angle δ of substation is considered as a reference.745 λ9 -31.15 -11. so the system contains (4+11+13-1=27) states: iT h iT h + ∆ω ∆λd ∆λf d ∆λkd ∆λq ∆λkq ∆δ 21 : ∆xgov ∆xval ∆xf p ∆ω ∆E 0 ∆E 0 q d + [∆xex ]T Table 4.24 -0.13 -6.4564 λ22 λ23.59 λ5 -22.CHAPTER 4.48i λ3.7 -31.4.956 λ8 -30.803 λ16 -1.4: Eigenvalues of the linearized distribution system with DGs. Eigenvalues -140.46964 λ27 h ∆ω ∆λd ∆λf d ∆λkd ∆λq ∆λkq ∆δ 31 : ∆xcon1 ∆xcon2 ∆xT F 1 ∆xT F 2 ∆xint iT + + [∆xex ]T The eigenvalues of the distribution system under the speciﬁc operating point are given in . 4 states of substation: ∆ω ∆E 0 ∆E 0 ∆δ q d 36 11 states of diesel engine synchronous generator + 1 integrator + 1 state of excitation system: [∆xex ] T 10 states of gas turbine synchronous generator + 1 state of excitation system: iT h + [∆xex ]T ∆ω ∆λd ∆λf d ∆λkd ∆λq ∆λkq ∆δ : ∆xgov ∆xval ∆xf p h ∆ω ∆λd ∆λf d ∆λkd ∆λq ∆λkq ∆δ : ∆xcon1 ∆xcon2 ∆xT F 1 ∆xT F 2 ∆xint iT + In the distribution system with substation and DGs.2 -30.73748i λ12. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK iT h .

.21) ∂λ2 ∂λ2 2 k = 1. D]. vi 6= 0 T T T wi A = λi wi ..5. r)..2 Eigenvalue Sensitivity In a linear system [A. .24) So. ∂λn ] ∂a ∂a ∂ak 1 2 = diag[ ∂a2 ... ... . r The terms ∂Λ ∂ak and ∂2Λ ∂a2 k in equation 4.26) λ1 . ∂λn ] ∂a2 k k k 1 ∂2Λ ∂Λ (∆ak ) + (∆ak )2 ∂ak 2 ∂a2 k (4.. 2.. ∆λn ] ∂Λ ∂ak ∂2Λ ∂a2 k 1 2 = diag[ ∂λk .25) (4... A second order Taylor-series expansion is given to estimate the ∆Λ over the parameter ak as: ∆Λ = where ∆Λ = diag[∆λ1 ... 2. C...CHAPTER 4. An important stability issue of the system is to determine the eigenvalues deviation (∆Λ ) corresponding to the change of the parameters ak .. if the system A has n diﬀerent eigenvalues. wi 6= 0 (4. r). 0 otherwise (4. if i = j. ∆λ2 . T The right (vi ) and left (wi ) eigenvectors of the system plant A (n × n) corresponding to the eigenvalue λi (i = 1. B.23) where λi is the ith eigenvalue.. assume the system plant A(n × n) is a function of a general system parameter ak (k = 1.. λn T W T : n × n matrix which has wi as its ith row V : n × n matrix which has vi as its ith column I: n × n identity matrix ... . n) are deﬁned by: Avi = vi λi . 2.. ∂a2 .. . DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK 37 4.21 are deﬁned as the ﬁrst-order and second-order eigenvalue sensitivities with respect to the system parameter ak (k = 1. 2. the following relations exist: Λ = W T AV I = WTV where Λ: diag h i (4.. ∂λk . . ..22) (4. . and the product of left and right eigenvectors has the characteristic: T wi vi = 1..

13 -1.27) (4.00171 λ27 T : denotes transposition Taking derivatives of both sides of equation (4.06∓921.0904i λ17. Xd -292.29) The derivative of equation (4.44i λ3.18 0.24 -0.00015 ∓ 0.00026i λ12.2170 λ8 14.00044i λ6.5: Eigenvalue sensitivity with respect to Xd .1141 ∓ 8.15 -14.022 λ21 0.31) .1231i λ14.01005 λ26 -0. Eigenvalues sensitivity vs.0002 λ5 -0.2 -4.30) ∂V V = 0 and W T ∂ak = 0.CHAPTER 4. NO.7 5. DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK 38 Table 4.0065 λ11 -0.7684∓101.105 λ16 0.28i λ1.458 ∓ 3.26) is: ∂V ∂W T V + WT =0 ∂ak ∂ak Using the relationship ∂W T ∂ak (4. the equation (4.20 14.974i)×10−6 λ23.0458 ± 0.00219 λ25 0.00056 ± 0.0064 λ22 (-0.4 -0.28) (4.25) with respect to a speciﬁc parameter αk yields: ∂Λ ∂ak = = = ∂W T ∂A ∂V AV + (W T ) V + (W T A) ∂ak ∂ak ∂ak ∂V ∂A ∂W T (W T )−1 Λ + (W T ) V + ΛV −1 ∂ak ∂ak ∂ak T ∂W ∂V ∂A V Λ + (W T ) V + ΛW T ∂ak ∂ak ∂ak (4.6977 λ10 0.0125i λ19.999 λ9 -0.29) can be written as: ∂Λ ∂A = (W T ) V ∂ak ∂ak (4.1963 ∓ 0.

The substation model for dynamic studies is developed. the parameter is changed from Xd 1.02. With ∂A A∆ak − A = ∂ak ∆ak (4.82 and ∆ak is 0. For example. 4. .32) where A∆ak is the system plant of the linearized system in Simulink with very small deviation of parameter ak . DGS SIMULATION AND LINEAR ANALYSIS WITH SIMULINK In conventional solution.8 to 1. the ∂A ∂ak is derived from the linearized ∂A ∂ak can be calculated as: 39 equations [69][73]. the the help of the Simulink toolbox. the eigenvalue sensitivity with respect to the parameter Xd is listed in Table 4.6 Summary The matrix/vector concept in Simulink is used to simulate the full-order synchronous generator.5. This method can simplify the simulation and avoid the algebra loop. It consists of a large inertia synchronous generator and a transformer.CHAPTER 4. While calculating the A∆ak .

A phosphoric acid fuel cell contains the primary subsystems shown in Figure 5. operating data on a phosphoric acid fuel cell system [66][83] was obtained.1.40 Chapter 5 A Phosphoric Acid Fuel Cell Model from Data Fitting 5. With the assistance of local electric utilities. The impedance characteristics of the fuel cell are investigated. the data on the fuel cell operation is retrieved and applied to obtain generic structures describing the dynamics such as transfer functions and their time domain counterparts.1 Introduction This chapter presents a fuel cell model obtained through data ﬁtting method with the operating data of a phosphoric acid fuel cell (PAFC). Dynamic behavior of components can be modeled using systems theory in several ways. The nonlinear simulation illustrates that the model is helpful to investigate the dynamic stability of the distribution system with a fuel cell power plant. diﬀerential equations. Dynamic simulations in MATLAB/SIMULINK are then studied toward the integrated system of the fuel cell for the IEEE 13-node power distribution system with existing gas turbine and diesel engine distributed generations (DGs). A large volume of data is available from recorded events on the subject fuel cell. . The impacts of fuel cell on the system are investigated. An eﬃcient way to develop the complicated fuel cell model is presented in this chapter. Here. Actual data was recorded by a Digital Fault Recorder (DFR) system monitoring the operation of the cell.

Thus the fuel cell can be represented as a controllable current source controlled with variable .c represent the double-layer capacitances near the anode and the cathode. A certain voltage is found as an open circuit voltage (OCV). Anode Electrolyte Cp.2 V-I Characteristics A PAFC can be electrically modeled as shown in the equivalent circuit of Figure 5.c corresponds to polarization resistance of the cathode.a Re Rp. The structure is composed of an electrolyte and two electrodes.a Rp.1: Diagram of a PAFC’s generalized schematic.a and Cp. Rp.2: An equivalent circuit of the PAFC. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 41 Hydrocarbon Fuel Oxygen (Atmospheric) Reformer H2 Fuel Cell Stack DC DC AC To Load/Grid Figure 5. Rp. CO2 H2O 5.CHAPTER 5.c Cp. The current increases with decreasing potential below the OCV.2 [135]. Re is the resistance of electrolyte in the matrix of the PAFC. The current-voltage nonlinear characteristics of a single PAFC with a variation in the amount of the dispersant can be obtained from the performance data. in parallel with the doublelayer capacitance of the interface between the electrolyte and the anode. The capacitive component of the electrolyte is ignored since the electrolyte is highly conductive.a represents the anodic polarization resistance. Cp.c Cathode Figure 5.

Figure 5. By dividing the voltage by the current the impedance is obtained. impedance. The RMS AC output voltage value is proportional to the DC output voltage value. The controllable current source and the output voltage can be obtained through data ﬁtting.CHAPTER 5.4: A model of the PAFC’s DC part. the voltage increases when the current decreases. (X axis: current (A).3 shows the I-V characteristic of the PAFC obtained from the recorded data during a shut down cycle process.3: I-V characteristics of the PAFC obtained from the recorded data. In the turning oﬀ process. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 42 280 260 240 220 200 180 160 140 120 100 -200 0 200 400 600 800 1000 1200 Figure 5. Therefore. Figure 5. a detailed fuel cell model can be constructed. Overall. Thus the whole fuel cell model is known. . if we can obtain the impedance characteristic of the fuel cell and the gas valve turn oﬀ process. The impedance characteristic is similar to that in [135]. the DC current is used to represent the valve of the fuel cell.4 shows the DC part of the PAFC model. Y axis: voltage (V)) Vout control signal I(t) Z(t) Figure 5.

1 Least Squares Optimization Algorithm The estimation of the parameter of the time domain function is accomplished through data ﬁtting. t) − φ(t))2 dt (5. Figure 5. The objective function is expressed as the integral of the least square error: min n Z t1 x∈R t2 (y(x. raw data.1) .CHAPTER 5. Identiﬁcation of the candidate time domain function.3 Data Fitting To obtain the transfer function H(s). 2. 5. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 43 Figure 5. Several such data samples were used to obtain the models in this chapter. 5. Estimation of the parameters in the time domain function.5 shows the sampled data of the DFR device.5: Diagram of the sampled data of the DFR device.3. Obtaining the transfer function from the actual time domain data. for the fuel cell from the analysis of the real. 3. Here the least square optimization method is used. there are three steps[134]: 1.

3) where k.2) where y and φ include the weights of the quadrature scheme.2.2 Fuel Cell Transient Output Current Modeling From the general form of the fuel cell’s DC current. The MATLAB Optimization toolbox was applied to solve this problem and generate the numerical optimal parameters. it can be seen that it has the general form of analytical response: I(t) = ke−α(t−t0 ) cos(ω(t − t0 )) + c (5.3) to the actual fuel cell data. The identiﬁed values of the parameters in equation (5. Figure 5.1 summarizes the best-ﬁt parameters for all three of the above calculated fuel cell shut down dynamics cases. A second approximation example (Case 2) selected was the DC current response on a diﬀerent data set (e. 5.1. Equation (5.3) are shown in Figure 5. t) is the estimated output function with x and t as parameters and φ(t) is the real data. t) − φ(t))2 i=1 x∈R (5. the equation (5. It can be seen that the parameter values given in Table 5.1) to the actual measured fuel cell data. Finally. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 44 where y(x.g. t0 and c are the undetermined constants. From Table 5. independent data set).3) parameter values are as given in Table 5.6 shows the DC current response corresponding to a fuel cell shut down (Case 1). one might select the typical set of parameter .3) approximation values (continuous line) and the equation (5.1 result in very good ﬁts of equation (5. The small circles denote the fuel cell (DFR) sampled data and the continuous line is the equation (5.3.7 gives the measured data (circles). They will be determined by ﬁtting this analytical expression to the current data record. Three diﬀerent data records were ﬁtted to this form.3) parameter values which resulted from the best ﬁt of equation (5.3) ﬁtted approximation of the actual data.1) is discretized and formulated as a least squares problem: min f (x) = n m X (y(x. y(x. Table 5. From these intervals of parameter values. the approximate ranges of equation (5. a third example fuel cell shut down data set (Case 3) is selected for approximation of the DC current. t) is required to follow the trajectory of φ(t). Figure 5.6.CHAPTER 5. ω. α. The data records from the DFR can be used to make the ﬁt using MATLAB.

5 Figure 5.6 0.3 1.9 time(sec) 1 1.9 1 1.044 α = 9.CHAPTER 5. 1200 1000 800 600 k = 775.3 Figure 5.1 1.467 t0=0.8 0.531 current (A) 400 200 0 -200 0.4 0.3699 t 0= 0.6206 α = 9.041 c =106.6: The recorded data and ﬁtted curve of the PAFC’s DC current (Case 1).2 1.1 time(sec) 1.6 0.5 0.83272 ω =14.7 0. .7: The recorded data and ﬁtted curve of the PAFC’s DC current (Case 2).2 1.7 0.8 0.0913 c =116 600 current (A) 400 200 0 -200 0.4 1.5080 ω = 14. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 45 1200 1000 800 k = 819.

CHAPTER 5. ω t0 14.3. k α c ω 750-819.25 115 14. The MATLAB function POLYFIT is used to match a polynomial in time to the fuel cell DC voltage data vs. the most appropriate analytical function form is a polynomial.0 Representative values shown in Table 5. ω = 14.0 9. MATLAB Optimization toolbox is used to ﬁt an approximation polynomial to actual fuel cell data provided.3 Fuel Cell Transient Output Voltage Modeling Next the fuel cell DC output voltage response is modeled. α = 9. 5. record data. In this case.04 0. Again.0 9. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 46 Table 5. the Laplace transform of the output response I(s) is: I(s) = ks c et0 s + (s + α)2 + ω 2 s (5. Considering the shut down to be a response to a unit step (or switched) input.6 9-9.5 110-120 13. POLYFIT adjusts .2: Range of parameters and representative parameter values. Taking the Laplace transform of equation (5. U (s) = − U0 s (5. c = 115.15-14.4) where U0 is the amplitude of the down step.9 9.508 14. time by minimizing the mean squared error between the analytical polynomial and the voltage data. the output current response to this input shut down is equation (5.25.045 111.533 Table 5.6) where k = 785.467 106.1: NO.6 9.3).370 116 775. NO.5 766.4 vs.87 0.0.09 0.15 Range 785.3).2 as representative values for a generic model of this class of PAFC fuel cell shut down dynamics.5) Therefore the transfer function is: H(s) = ks c I(s) =− et0 s − 2 + ω2 U (s) (s + α) s (5.833 13. Case 1 Case 2 Case 3 DC current parameters k α c 819.

95 7 4.9) As an example. aN −1 . N Error 4 29.70 5 9. polynomial order.12 9 1. . . a0 ] (5. aN−2 .0003 −0.3 and Figure 5.7) The root mean square error between any set of sample values of the fuel cell’s DC voltage data and equation (5. a1 .10) and the polynomial is: 8 X V (t) = n=0 coef (n)(t − t0 )n (5. A graph of this relationship plotting error on the vertical axis versus number of polynomial terms (the order of the polynomial) is shown in Figure 8.0e + 006 ∗ [ −0.9 shows the recorded data and ﬁtted curve of the output voltage.11) Figure 5.94 13 1.0285 0.2339 0.95 11 1.3: Error vs. The equations are: V (t) = The vector of coeﬃcients is: Coef = [aN . A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 47 Table 5.0002 ] (5.CHAPTER 5. From Table 5.8.7877 −0.3.8) N X n=0 an (t − t0 )n (5. Therefore only an eighth order polynomial is needed to represent this transient DC voltage.3581 .94 the polynomial coeﬃcients to achieve the best ﬁt. . . .10 8 2.94 12 1.03 6 5.4206 1. the value of the error versus the number of terms in the polynomial for polynomials of order four through thirteen are shown in Table 5.7) values is: qX (Vsampling − Vestimate )2 Error = (5. for the Case 1 fuel cell data set.96 10 1. The coeﬃcients of the 8th order polynomial are: Coef = 1.1891 −1.0000 0. it is observable that the 8th order polynomial is almost as good as any polynomial of higher order to approximate the DC voltage’s waveform.

4 0.9: The recorded data and ﬁtted curve of the PAFC’s DC voltage (case 1).3 0. (x axis: time (s).2 0.508 240 230 220 210 200 0 0. 270 260 8th order error=2.1169 250 t0=0.8: Diagram of the Error vs.1 0. polynomial order for the PAFC’s DC voltage. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 48 30 25 20 15 error 10 5 0 4 5 6 7 8 9 the order of polynomial 10 11 12 13 Figure 5.8 Figure 5.6 0.7 0.5 0. y axis: voltage (v)).CHAPTER 5. .

CHAPTER 5.10: Diagram of the proposed PAFC’s dynamic model. The models of the gas turbine synchronous generator and diesel . and the open circuit impedance is expressed as: Z(t) = V (t) I(t) (5.6). The diesel engine model consists of the control system. 5. and DC current (A) outputs from the fuel cell model are available in Matlab. The impedance curve of the fuel cell model is shown in Figure 5. In Figure 5. In Figure 5. The gas turbine model consists of speed governor. the synchronous generator connected with the gas turbine and the diesel engine is represented by a full-order model. Figure 5.10.13 shows test system to evaluate the impacts of the PAFC on the power distribution system. Z(t) in equation (5. the I(t) is determined by the H(s) in equation (5. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING Fuel Cell 49 Vout H(s) I(t) Z(t) DC/AC Figure 5. valve position.4 Fuel Cell Modeling Figure 5. and turbine. 5.13.3.12) can be calculated. actuator. The IEEE-13 node power distribution system with multiple types of DGs is represented in Matlab/Simulink by an average model.12. The PAFC model is connected to the gas turbine DG directly. The test system is the IEEE 13-node system with a gas turbine synchronous generator and a diesel engine synchronous generator.10 shows the proposed fuel cell model.11 gives the resulting V-I curve. integrator and the engine itself. fuel pump.12) Since the DC voltage (V). The model represents the turn down process of the fuel cell.4 Conﬁguration of Test System with DGs Figure 5.

3 0.1 0. . 1400 1200 1000 800 600 400 200 0 -200 0 0. (X axis: time (s). time. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 50 270 260 250 240 230 220 210 200 -100 0 100 200 300 400 500 600 700 800 900 Figure 5.11: The V-I curve of the obtained PAFC’s model.5 0.7 Figure 5.CHAPTER 5. (X: current (A). Y: voltage (V)).2 0.4 0.12: The impedance of the obtained PAFC’s model vs. Y axis: impedance (Ω)).6 0.

In this chapter.11 means the impedance of the PAFC can be maintained relatively constant within the operating range. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 51 Subst at ion 650 646 645 632 633 634 diesel -engine syn gen 611 684 671 692 675 652 680 PAFC gas-turbine syn gen Figure 5. This is a very important characteristic of the fuel cell power plant to implement the real/reactive power control via the DC/AC inverter. These base values are used in Chapters 5-10. The base of the distribution system: Sbase = 4. 5. Vbase = 4. The initial power penetration of gas turbine is Sgas = 0.5 Simulation Study In section 5. not with the Power system Blockset. the current input from the PAFC aﬀects the dynamic stability of the distribution system directly. The approximately monotonic change of the V-I curve in Figure 5. and the impedance characteristic curves are studied respectively.CHAPTER 5. The eﬀects of voltage are relatively indirectly. A simpliﬁed excitation system is used.0M V A. current curves.2625 + j0. During the turn oﬀ procedure. the voltage curves.1166 (pu).16kV . the current model of the PAFC is used to investigate the impacts of the PAFC on the dynamic stability of the power distribution system during PAFC turn oﬀ. the full-order synchronous generator is simulated by Simulink. the initial . engine synchronous generator have been described in detail in Chapter 4 [81]. Here.3.13: IEEE-13 node power distribution system with mult-types of DGs.

The model is useful to evaluate the dynamic impact of the PAFC power plant on the .17 show the simulation results to investigate the impacts of the fuel cell model on the distribution system.165.14-5. Figure 5. Two cases are conducted with diﬀerent initial real power penetrations from the fuel cell power plant: Case I: Pf uelcell = 0.15 is the bus voltage magnitude. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 52 Figure 5. 5. Figure 5.CHAPTER 5.14 shows the dynamic procedure of the real power penetration from the fuel cell power plant to power distribution system. Case II: Pf uelcell = 0. Figures 5.17 show the dynamic response of the gas turbine and diesel engine DGs.02475.0825. dotted lines represent the Case I and solid lines represent the Case II.6 Summary This chapter presents a practical method to develop the complicated fuel cell model via the data ﬁtting.1026. In the ﬁgures. power penetration of diesel engine is Sdiesel = 0.1375 + j0.14: Real power penetration from the PAFC power plant to IEEE-13 node power distribution system (dotted line: Case I. The simulation results show that the larger penetration of fuel cell will produce great impact on the system during the turn-oﬀ procedure. solid line: Case II). Figures 5.

solid line: Case II). Figure 5.15: Voltage magnitude of bus 671 during the PAFC’s turn-oﬀ procedure (dotted line: Case I.16: Rotor speed response of the gas-turbine synchronous generator during the PAFC’s turn-oﬀ procedure (dotted line: Case I.CHAPTER 5. A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 53 Figure 5. solid line: Case II). .

solid line: Case II).17: Rotor speed response of the diesel-engine synchronous generator during the PAFC’s turn-oﬀ procedure (dotted line: Case I. The obtained impedance characteristic of the fuel cell model is the precondition to develop the inverter-based controllable fuel cell model. The nonlinear simulation in Matlab/Simulink was done to investigate the impacts of the PAFC model on the power distribution system with diﬀerent power penetration. The results show that the larger penetration of fuel cell will produce greater impact on the system during the turn-oﬀ procedure.CHAPTER 5. power distribution system. . A PHOSPHORIC ACID FUEL CELL MODEL FROM DATA FITTING 54 Figure 5.

In the developed model (see Figure. Besides the control of the fuel supply. analyze each component of a SOFC power plant system. These characteristics of fuel cell are attracting increasing power provider’s interest. The objectives of this chapter are to: 1.1). Normally.6. A fuel cell power plant is a type of DG to convert chemical energy into electricity via the electrochemical procedure directly. less noise and lower emissions. To study the impact of the fuel cell power plant on the power distribution. the conventional electric generation. a controllable dynamic model of a SOFC power plant is developed in this chapter. Base on the model in [92] (see Figure 2.55 Chapter 6 Dynamic Modeling of A Solid Oxide Fuel Cell Power Plant 6. the connection between the SOFC and the power conditioning unit (PCU) is equivalently represented as the voltage (vf c ) instead of the real power.1 Introduction In Chapter 5. the dynamic model and stability control of solid oxide fuel cell (SOFC) power plant are investigated. a methodology is presented to obtain a phosphoric acid fuel cell (PAFC) model at certain operating condition. the model is simpliﬁed and developed to be suitable for simulation in Simulink. determine the system structure . However. the adjustment of real power transmission between the SOFC and PCU will be implemented by the voltage. the electrochemical process of fuel cell is lower temperature. In this chapter and Chapter 7. such as gas turbine and diesel engine DGs convert the energy via the rotating generator.2).

the real and reactive power set point (P ∗ . 6. in which the plant is grouped into ﬁve parts: fuel cell stack. of a typical fuel cell power plant. A plant structure of a SOFC based DG is presented in [92]. measured voltage (vd . DYNAMIC MODELING OF A SOLID OXIDE FUEL CELL POWER PLANT 56 id. are used to obtain voltage (Vf c ) as the output of the SOFC. network interface controller and plant controller (see Figure 2. power conditioning unit.iq grid set P*. simulate the SOFC power plant in Simulink.1.1: Diagram of the general structure of a SOFC power plant. 2. 3.vq id.2).Q* Controller measure Figure 6. In the Figure 6. develop the dynamic model of the SOFC power plant and represent it as ordinary diﬀerential equations.2 Structure of A SOFC Power Plant Basically.iq* vd. The output of a fuel cell is DC current which will be transformed to AC currents through dc/ac inverter to be connected with ac network. balance of plant. .1) from [92].iq fuel air P* Fuel cell (SOFC) vfc Ifc PCU (CSI) id*. and test the dynamic response of the model. iq ).CHAPTER 6. Q∗ ). The structure of a controllable dynamic SOFC power plant is developed (see Figure 6. vq ) and current (id . fuel cells are devices that transform chemical energy into electrical energy.

tH2O . The amount of the fuel and air is set by the set real power (P ∗ ). the voltage of a fuel cell is determined by the three state variables: partial pressure of hydrogen. partial pressure of oxygen and partial pressure of steam. DYNAMIC MODELING OF A SOLID OXIDE FUEL CELL POWER PLANT 57 load + air Fuel H2O Anode Cathode Figure 6.2: Structure of a fuel cell. The terminal voltage of the fuel cell plant depends on the number of in series connected cells and the capacity is determined by the total number of the cells. . the size of the load aﬀects the terminal voltage of the cell. A practical fuel cell power plant consists of one or more fuel cell stacks. Each fuel cell stack has many seriesly connected cells. the absolute temperature. The time constant tH2 . the universal gas constant. In this chapter.3) N0 2F . and an electrolyte (see Figure 6. and so on. Also. Kr = N0 is the number of cells of one stack.CHAPTER 6. a cathode. From the Nernst’s law.2). The partial pressure of the three type of substances in a cell can be expressed as: pPH2 = − 1 tH2 (PH2 + 1 tH2 O 1 (q in − 2Kr If c )) KH2 H2 2Kr If c ) KH2 O (6. tO2 are determined by the volume of the anode. the dynamic model of SOFC [92] is presented. 6.2) (6.3 Dynamic Model of Solid Oxide Fuel Cell (SOFC) A fuel cell consists of an anode. F is the Faraday’s constant. pPH2 O = − pPO2 = − 1 tO2 (PH2 O + (PO2 + 1 (q in − Kr If c )) KO2 O2 in in where qH2 and qO2 are the molar ﬂows of hydrogen and oxygen. The output of the cell is the direct current voltage (Vf c ).1) (6. Here. one fuel cell stack is simulated and studied.

5 PH2 PO2 RT (ln( )) − rIf c 2F PH2 O = VNernst − rIf c (6.4) (6.CHAPTER 6.3. DYNAMIC MODELING OF A SOLID OXIDE FUEL CELL POWER PLANT 58 Inverter + Vfc k. output voltage/current modulation.3: Three-phase CSI-based PCU. The following equations show the voltage of the fuel cell stack: Vf c = N0 (E0 + 0. The input of inverter are the pulse width modulation (PWM) gain k and angle α.5) where If c is the current of the fuel cell stack.3. Rs represents the sum of the losses of a Y /∆ transformer resistance and the inverter conduction. Ls represents the leakage inductance of the transformer. . Three-phase equivalent circuit of PCU is shown in Figure 6. 6. The DC voltage Vf c is the output of a SOFC. α Current controller Rs Ls ia eb ec ib ic vsb vsc ea vsa to grid measurement Figure 6. In Figure 6.4 Power Conditioning Unit (PCU) A PCU is an interface to connect the fuel cell with the utility grid. ic are the three-phase AC current output of inverter. eb . To study the dynamic characteristic of PCU. The voltage of the fuel cell stack can be determined by the Nernst’s equation and Ohm’s law. ec are the three-phase AC voltage output of inverter. a dynamic model of PCU for fuel cell is developed. The state equations of the three-phase current source inverter (CSI) based PCU are the diﬀerential equations of voltage and current [101][132]. The basic functions can be summarize as: DC/AC transformation. ib . and ia . ea .

6) is transformed to the system synchronous reference frame as: i pi q = AI q + BI1 Vf c + BI2 Vs pid id where AI = s − RLωs 0 s 0 s − RLωs 0 s (6.CHAPTER 6.1).6) The phase a grid voltage at the interconnecting point PCU is given by: vsa = √ 2Vs cos(ω s t + θs ) (6. BI1 = BI2 = ωs ωs L0 k sin(α + θ s ) s ωs 0 k cos(α + θ s ) Ls ω − Ls sin θs 0 .5 Inverter Control Unit Besides the fuel cell (SOFC) and inverter. DYNAMIC MODELING OF A SOLID OXIDE FUEL CELL POWER PLANT 59 pip = − where p = {a.7) where Vs ∠θs is voltage phasor of connecting bus. which represents the interaction between inverter and fuel If c = 1 0 Vf c + id k cos(α + θ s ) + iq k sin(α + θ s ) Rdc (6.10) 0 where Rdc represents the switching losses.8) (6. b. c}. the state equation (6. For the sake of stability analysis and controller design. The current tracking inverter control strategy can be used . the relationship between voltage and current may be expressed as: . the control unit is another important component in the fuel cell power plant (see Figure 6. 6. and ω s is the system frequency. which represents the interaction between inverter and grid. s .9) −ω s 0 cell. ω − Ls cos θs 0 s In the DC circuit. The ac side phase a voltage of inverter is: ea = kVf c cos(ω s t + α) where k and α are the PWM modulation gain and angle. Rs 1 ip + (ep − vsp ) Ls Ls (6.

i∗ . Thus the fuel cell power plant can be considered as a PQ bus or a PV bus depending on the diﬀerent control strategies.12) p∆id ∆id ∆vd ∆α where A∆ = 0 s − RLω0 0 s s − RLω0 0 s −ω 0 0 B∆1 = ω0 − ω0 L0 s − ω0 L0 s . the equation (6.4: Diagram of the current control scheme for the CSI-based PCU. The basic idea of current tracking inverter control is illustrated in Figure 6. The input values are errors between measured id .4.CHAPTER 6. or to track the real power (P ∗ ) and voltage (V ∗ ) requirement.9) is linearized under the speciﬁc operating point as [82]: ∆k = A∆ + B∆1 + B∆2 ∆Vf c p∆id ∆id ∆vd ∆θ p∆iq ∆iq ∆vq ∆α (6. The output of the switch-mode inverter is the AC voltage v cos α (kVf c cos(ω s t + α)). which has output k. DYNAMIC MODELING OF A SOLID OXIDE FUEL CELL POWER PLANT 60 Comparator id*. the current tracking PQ type control strategy is presented. Here. the linearized model can be simpliﬁed as: ∆iq ∆vq ∆k p∆iq = A∆ + B∆1 + B∆2 (6.11) Considering that the DC voltage deviation of the SOFC is very small ∆Vf c = 0. to track the real power (P ∗ ) and reactive power (Q∗ ) requirement.iq* + Σ id. the voltage deviation of the grid is very small ∆θ = 0. iq and reference i∗ .iq PI control SwitchMode Inverter vcosα Figure 6. . To develop the dynamic model of PCU. α. They are passed through to the PI d q controller and comparator. The relationship between the inverter controller parameters and the current has been expressed in (6.9).

Figure (6.8) show the measured P and Q responses and Figure 6.7 illus- .12). Rs = 0. DYNAMIC MODELING OF A SOLID OXIDE FUEL CELL POWER PLANT 61 from measuring iq PI vq from measuring vd p* set point q* i*q calculate unit i*d + + PI id from measuring Σ PI - + + Σ ∆k to inverter Σ PI + + Σ ∆α Figure 6. the calculate unit represents the following equations. the current id and iq can be controlled by the ∆ = k11 k12 adjustment of the gain k and/or ﬁring angle α.81e−4 . ω 0 Vf c L0 s ω 0 Vf c L0 s ω 0 Vf c L0 k0 cos(α0 + θ s ) s ω0 Vf c − L0 k0 sin(α0 + θs ) B∆2 = Figure 6.7-6. the fuel cell power plant is simulated in Simulink. Figure 6.01. KO2 = 2.875 to 0.126. The single fuel cell power plant and inﬁnite bus system (see Figure 6.52e−3 . Ls = 0.6 Model Simulation and Test To test the mathematical model presented in previous sections.7 at t=5.1.0s and the reactive power set point Q* remains constant. sin(α0 + θs ) cos(α0 + θs ) k21 k22 s According to the linearized equation (6. i∗ q ∗ id vd v 1 q 2 + vd −vd vq Q∗ P∗ = 2 vq (6.43e−4 .13) 6. KH2 = 8. tH2 = 26.5: Diagram of an average model of a CSI-based PCU.CHAPTER 6. One case study is conducted to test dynamic response as the real power set point is changed. r = 0.1s.9 shows the terminal voltage response of SOFC.3. The delay time of the inverter is 0.5. tH2O = 78. tO2 = 2.91. The simulation results illustrate the capacity to track the real power set point.6) are used to test the eﬀectiveness of the developed model.5. The parameters of fuel cell [92] are: N0 = 384. KH2O = 2.9. The real power set point P* makes a step change from 0. Then the control unit of CSI is demonstrated in In Figure 6.

a dynamic SOFC power plant model is developed.1s. 6.75 in 0. Q Figure 6. The SOFC internal Nernst’s voltage is determined by the three state variables inside the SOFC. Figure 6. As a contribution of this DG research. trate the real power changes from 0.875 to 0. an average PCU model is proposed with four PI controllers to connect the SOFC into the distribution system.Q* Fuel Cell Power Plant infinite bus measuring P.7 Summary In this chapter.6: Diagram of single fuel cell power plant and inﬁnite bus. .10 shows the SOFC internal Nernst’s voltage response of the step change in real power set point. The nonlinear simulation results illustrate that the SOFC power plant can track the real/reactive power set points rapidly.CHAPTER 6. The voltage speed is very slow due to the large time constants of the pressures. The response speed slows down due to the change of the SOFC terminal voltage. DYNAMIC MODELING OF A SOLID OXIDE FUEL CELL POWER PLANT 62 set P*.

8: Reactive power response of the step change in real power set point.7: Real power response of the step change in real power set point. Figure 6.CHAPTER 6. DYNAMIC MODELING OF A SOLID OXIDE FUEL CELL POWER PLANT 63 Figure 6. .

CHAPTER 6. . Figure 6.10: SOFC internal Nernst’s voltage response of the step change in real power set point. DYNAMIC MODELING OF A SOLID OXIDE FUEL CELL POWER PLANT 64 Figure 6.9: SOFC terminal voltage response of the step change in the real power set point.

1 Introduction The fuel cell power plant is gradually used as a new type of distributed generation (DG). 7. The dynamic model of a SOFC power plant studied in Chapter 6 illustrates that the fuel cell power plant can track the real and reactive power instantaneously. the integration of a SOFC power plant dynamic model to the power distribution system is investigated and the impact of the fuel cell on the system is studied.1s. the linearized model of the system with a fuel cell power plant. the fuel cell power plant is connected to the IEEE-13 node distribution system with a gas turbine synchronous generator and a diesel engine synchronous generator (see Figure 7. a gas turbine DG and a diesel engine DG is developed. The fuel cell power plant and the gas turbine DG are connected . Based on the linearized model. These rapid real/reactive power track characteristics provide the potential of a fuel cell power plant to improve oscillation damping in the power distribution system.1). The average model of the IEEE-13 system with a gas turbine DG and a diesel engine DG has been presented in Chapter 4 [81].2 System Conﬁguration To study the impact of the fuel cell on the power distribution system with multiple types of DGs. The response time of the inverter is 0.65 Chapter 7 Dynamic Stability Control for A SOFC Power Plant 7. the optimal control is designed and the controller is tested under diﬀerent fault locations. In this chapter.

the linearized model plays a very important role to analyze the nonlinear system. To evaluate the impact of the fuel cell power plant on the distribution system.05. the fuel cell supplies the power s = 0. Pdiesel = 0. DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 66 Subst at ion 650 646 645 632 633 634 diesel -engine syn gen (DG2) 611 684 671 692 675 inverter 652 680 gas-turbine syn gen (DG1) SOFC fuel cell power plant (DG3) Figure 7.0875.1). In the following sections.1578.2 shows the ground fault disturbance results in the rotor speed deviation of DG1 and DG2.4) show the simulation results.0875.CHAPTER 7. The parameters of the substation. 7.2625 + j0. Figure 7. based on which controllers are designed to improve the oscillation damping of the distribution . and Psof c = 0. The operating condition of the power distribution system with DGs is listed as: Pgas = 0. the integrated system can be represented by linear models. at the same bus. the gas turbine DG and the diesel engine DG are listed in Appedix B (Table B. the controller is designed to improve the oscillation damping.2-7. The Matlab/Simulink provides a powerful tool to linearize the nonlinear system in Simulink under a speciﬁc operating condition. Figures (7.0. Around that operating point.3 System Linearization As discussed in Chapter 4.0. a case study is conducted that a ground fault occurs at bus 632 and this fault is cleared after 0. At the initial condition. Vgas = 1.1: IEEE-13 node power distribution system with multi-types of DGs to evaluate the impact of a SOFC power plant. Vdiesel = 1.25s.

DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 67 Figure 7.3: Real and reactive power of the SOFC power plant as a fault occures at bus 632 without controller. . Figure 7.2: Rotor speeds of DGs 1&2 as a fault occures at bus 632 without controller.CHAPTER 7.

the 3rd order gas turbine h iT and the 1st order excitation system: ω ir ir d ir ir ir δ : xgov xval xf p : xext q d f kd kq Diesel engine DG consists of the 7th order synchronous generator. the linearized models of the distribution system with multiple types of DGs are obtained by the Simulink. To analyze the dynamic characteristics and design the dynamic controller of the distribution system with DGs. h iT ω ir ir d ir ir ir δ : xcon1 xcon2 xT F 1 xT F 2 xint : xext q d f kd kq iT h 0 0 Substation is represented as the transient synchronous generator model.4: Terminal and internal Nernst voltage of SOFC as a fault occures at bus 632 without controller. 2nd order dynamic PCU model and four The state variables x = of linearized system. the 5th order diesel engine and the 1st order excitation system. Gas turbine DG consists of the 7th order synchronous generator. The linearized models of a gas turbine DG. a diesel engine DG and the substation have been presented in Chapter 4. DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 68 Figure 7.CHAPTER 7. ω Eq Ed δ The fuel cell consists of 3rd order SOFC model. h iT PI controllers: PH2 PH2 O PO2 : iq id : xP I1 xP I2 xP I3 xP I4 The four integral states of the PI controller are weakly connected to the oscillation modes . system.

5. To select the control signal. The dominant eigenvalues are listed in Table 7. Based on the linearized model (A.6650i λ21. The controller design is studied in the following section. NO. the linearized power distribution system has 36 total state variables. especially modes (λ21.1) (7.1: Dominant eigenvalues of the linearized distribution system with 3 DGs.4 Controller Design to Assign Eigenvalues Consider the above linearized power distribution system with multiple types of DGs represented in the form: x = Ax + Bu ˙ y = Cx where x are the linearized state variables vector and u is the input of fuel cell. the participation factor is used to select the (7.1.5: Structure of the linearized output feedback control.1168 ±9.22 -1. the controller is designed via eigenvalue analysis. The controller objective is to improve the oscillation damping of the system.26 ). λ25. DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 69 u0 + - dx/dt=Ax+Bu x C y Controller Figure 7. in the linearized distribution system. C. Open Loop Eigenvalues -0.2) . The linearization results in Simulink verify above analysis. B. D). Table 7. there are in total 36 state variables.3.9580 ± 4. From the analysis in section 7.22 . only two rotor angles are independent.26 Once the three synchronous generators (2 DGs + Substation) are connected in the distribution system. 7. The structure of the linearized output feedback controller is shown in Figure 7.CHAPTER 7. So.0009i λ25.

the linearized output feedback control is designed.6. is deﬁned. The four participation factors of each oscillation modes are very close.2.2363 ∆δ gas_substation 0. The dimension of PM is same as system matrix A.26 ). The centralized control scheme is proposed in Figure 7. and the element PM(k.26 0. i) of PM is |P Fki |. PM. The lead-lag network is designed based on the residue of each input control signal respectively. The participation factor P Fki reﬂects the participation of the kth eigenvalue in the zero input response of the ith state variable. D). each of the 36 states has a factor related to the modes (λ21. DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 70 Table 7. A large factor |P Fki | in PM means the eigenvalue λk is mostly associated with xi . k2) including the phase compensation unit. State Variables PF of λ21.2267 0. the four outputs of the system may be considered as the input of the controller. λ25.26 . In the linearized distribution system.P Fki . The phase compensate unit is the lead-lag compensation network. For the sake of the convenient measuring and signal communicating.2398 0. .22 . and the rotor angle diﬀerence between the diesel engine and the substation. The participation factors demonstrate that the most relative state variables are: rotor speed of the gas turbine DG. To improve the oscillation damping of the dominant modes. C. To associate a state variable with an eigenvalue.2668 ∆δ diesel_substation most relative state variables to the critical modes. The participation factor. The four most related states are listed in Table 7.CHAPTER 7. B.2: Participation factors of eigenvalues λ21. D) is deﬁned as: P Fki = Vik Wki (7.2483 0. the rotor speed of gas turbine DG and diesel engine DG are selected as the control signal. λ25.2420 0. C. In the controller design. The output optimal feedback control algorithm is implemented to select the centralized gains (k1.3) where V and W are the right and left eigenvectors of the system (A. the participation matrix. the rotor angle diﬀerence between the gas turbine DG.2684 ∆ω diesel 0.22 . rotor speed of the diesel engine DG. B. of system (A.22 PF of λ25.2273 ∆ωgas 0.

the coordinated control scheme has two inputs and one output. K(s) = ( which consists of m phase lead-lag stages. 7. In each phase lead-lag stage. Time constant tj is evaluated from: τj = 1 √ ωj αj (7. Thus. and Rl.4) where l = 1. λ25. ∆ω diesel ) are considered to compensate the two pairs of the critical modes (λ21. + s − λ1 s − λ2 s − λn (7. 2 for each input.26 ) respectively.1 Rl.1 Phase Angle Compensation In Figure 7. each of the SISO system can be described as is: Gl (s) = Rl.22 .6) 1 + αj τ j s m ) 1 + τ js (7.2 Rl.. . The control inputs (∆ω gas .2p* phase compensate k2 gain P* Fuel Cell Power Plant Figure 7. two single input single output (SISO) systems are considered. αj is determined by the maximal compensation angle of this stage at frequency ω j : φj (ω j ) = sin−1 ( αj − 1 ) αj − 1 (7. The least damped critical modes should be considered ﬁrst.4.2p* ∑ Controller of Fuel Cell ∆ωdiesel -0.6: Coordinated control scheme of the SOFC power plant..7) At this time constant.n + + . the maximum phase lead-lag occurs at frequency ωj . DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 71 ∆ωgas phase compensate k1 gain ∆p ∑ 0.5) where ωj can be the imaginary part of a critical mode or a frequency where phase margin is needed.i is the residue of the eigenvalue λi .6. The lead-lag network shifts the phase of input signal from the plant G(s)[117]. If the linearized system has n distinct eigenvalues.CHAPTER 7. The lead-lag network is used to compensate the critical modes. thus to improve the oscillation damping of the power distribution system.

001877i = 0.9) The residues show the phase angle can be compensated up to 70. k2 ) of the feedback controller. τ = 0.11) (7. Consider the linear system represented by equations (7.22) = 0.0◦ . The corresponding residues are: R1.15) (7.26 with ∆ωdiesel The signal ∆ω diesel is considered to improve the oscillation damping of the modes λ25.6-7.1-7.26 .10) (7. the parameters of one compensation unit are: α = 3.26) = 0. The optimal gain design equations: 0 = AT M + M Ac + C T K T RKC + Q c 0 = SAT + AC S + X C K = RT B T M SC T (CSC T )−1 (7.74◦ The residues show that there is no need to compensate the modes(λ21.22 with ∆ωgas 72 To compensate the phase angle of ∆ω gas . To compensate the 70.CHAPTER 7.0118 ± 0. The corresponding residues are: R2.12) (xT Qx + uT Ru)dt (7.69.7).2 Optimal Control Algorithm to Select Gains The optimal control algorithm is used to determine the gain (k1 .95◦ (7. 7.22 ). From the equations (7.22 .95◦ .14) (7.05784.8) (7.00119i = 0.4.0118∠ ± 5.13) 0 with Q = QT > 0 and R = RT > 0.01986∠ ± 70.2). 2) Compensate λ25.(25. two stages of lead-lag networks are designed.16) . The output feedback controller is: u = −Ky Deﬁne the performance index: J= Z ∞ (7. DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 1) Compensate λ21.(21. this signal is considered to improve the oscillation damping of the modes λ21.0006483 ± 0. each stage contributing 35.95◦ total phase angle.

26 Table 7.17) In the linearized distribution system (A. ∆ω diesel ) of gas turbine DG and diesel engine DG are considered for the initial condition. the dimension of B is 38 × 1.869i -1.5T race(M u) (7. C. In the ﬁgures.298 ±8. The closed-loop system dominant eigenvalues are listed in Table 7. Simulation results show that the control via the fuel cell can eﬀectively improve the oscillation damping of the two synchronous generator DGs. the .3 illustrates the gains with/without phase angle compensation through the optimal feedback control algorithm. With phase Compensation Without phase compensation -53.1168 ±9. The nonlinear simulation results with/without controllers are shown in the following section.328 ± 4. The performance index is: J = 0. To compute the optimal gain (k1 .15 show the simulation results.4: Dominant eigenvalues of the open/closed loop system. DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 73 Table 7. B. λ21. Table 7.22 -37. In the performance index.22 λ25.56 k1 -57. Figures 7. The cases were done under the condition that a ground fault occurs. the Q is 38 × 38 identity matrix and R = 10. and the dimension of D is 2 × 1.4.35 -76.075i © ª where Ac = A − BKC.562i -1.CHAPTER 7. D). the dimension of A is 38 × 38. the dimension of C is 2 × 38. and initial condition of other linearized state variables are set as zeros. and X = E x(0)xT (0) .3: Feedback gains through the optimal control algorithm. the deviation of rotor speed (∆ω gas .5 Simulation Results In this section the dynamic simulation results are presented with/without controller. The operating condition was described in the previous section. 7. k2 ). the dashed lines represent the dynamic response without controller. controller #2 is output feedback controller with compensation unit. Two types of controller are tested: controller #1 is output feedback controller without phase angle compensation.6650i -1.868i -2.0009i -1. Closed loop Eigenvalues Closed loop Eigenvalues open loop eigenvalues with compensator without compensator -0.46 k2 Iterations 523 498 NO.699 ± 3.7-7.159 ±9.9580 ± 4.

dotted lines represent dynamic response with controller #1. Three cases with diﬀerent fault locations are simulated to test the robustness of the optimal output feedback controller. The fault occurs closer to the gas turbine DG. The simulation results verify that the optimal controller can eﬀectively improve oscillation damping of the power distribution system with DGs.6 Summary In this chapter. The simulation results show that the fuel cell with appropriate controllers can eﬀectively improve the oscillation damping of the whole system.9 show the simulation results.7-7. and the solid lines represent dynamic response with controller #2. Case 2: Fault occurs at bus 684. Case 1: Fault occurs at bus 634. DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 74 Figure 7. Multi-variable output feedback optimal controller based on a linearized model is designed using lead-lag compensator and optimal control algorithm. Case 3: Fault occurs at bus 632.7: Rotor speed of the gas turbine DG when a fault occurs at bus 634 (Case 1). The fault occurs between the diesel engine DG and the gas turbine DG. Figures 7.15 show the simulation results.10-7. 7. Figures 7. Figures 7. . The fault occurs closer to the diesel engine DG.CHAPTER 7. the impacts of the dynamic SOFC power plant on a distribution system with multiple DGs are investigated.13-7.12 show the simulation results.

CHAPTER 7.9: Terminal voltage of the SOFC when a fault occurs at bus 634 (Case 1). Figure 7.8: Rotor speed of the diesel engine DG when a fault occurs at bus 634 (Case 1). . DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 75 Figure 7.

DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 76 Figure 7.11: Rotor speed of the diesel engine DG when a fault occurs at bus 684 (Case 2).CHAPTER 7. Figure 7. .10: Rotor speed of the gas turbine DG when a fault occurs at bus 684 (Case 2).

. Figure 7.CHAPTER 7.13: Rotor speed of the gas turbine DG when a fault occurs at bus 632 (Case 2). DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 77 Figure 7.12: Terminal voltage of the SOFC when a fault occurs at bus 684 (Case 2).

DYNAMIC STABILITY CONTROL FOR A SOFC POWER PLANT 78 Figure 7.CHAPTER 7. .15: Terminal voltage of the SOFC when a fault occurs at bus 632 (Case 3).14: Rotor speed of the diesel engine DG when a fault occurs at bus 632 (Case 3). Figure 7.

The ma- . Currently. higher reliability and simpler control system. sub-rated power regime and rated power regime.79 Chapter 8 Model of the Wind Turbine DG 8. variable blade pitch angle control are installed[1][68][88]. a dynamic induction generator model with control requirement is studied and the possibility of contributing negative and evaluation of its eﬀectiveness in application in a power system is described. However. the voltage of isolated unit is regulated by the SVC and oscillation damping is improved in two ways: through multi-state feedback and through multi-output feedback. wind energy also has some limiting characteristics such as: unschedulable. the operating condition of the wind turbine is classiﬁed into several regimes: startup regime. a wind speed of 29MPH is needed for the power output to be rated. A wind turbine synchronous generator unit model is given in [68][88]. In [1]. Dynamic simulation and the controller design are investigated in this regime. A wind turbine induction generator unit model was developed in [1]. uncontrollable. To obtain relatively constant power. etc. In this chapter. A detailed dynamic turbine model was presented in [88]. the induction generator is attracting interest because of its lower cost.1 Introduction This chapter and Chapter 9 focus on the model and dynamic stability control of the wind turbine DG. The blade pitch control is used to achieve synchronous rotor speed[88]. In [88]. In rated power regime. The wind turbine induction generator is an attractive distributed generation (DG) in a deregulated electric energy environment since wind energy is a non-polluting source. a wind turbine powered induction generator is simulated. In [79].

8. Our research focuses on the dynamic performance of the wind turbine in the rated power regime. an induction generator and a SVC.2 System Models The wind turbine induction generator.8m/s. When the speed of the wind is greater than 6.CHAPTER 8. 8.1) . the wind turbine works in the rated power regime. The dynamic output mechanical torque of the wind turbine is expressed [1][88]as: 1 2 Tm = ρARCp Vω /λ 2 (8. The ﬁxed capacitor and thyristor controlled reactance (FC-TCR). Once the speed of the wind reaches 12.2. is used to compensate the var of the induction generator. The region between start-up and rated power regime is called the sub-rated power regime.1 Wind Turbine Model Based on the speed of the wind. the start up procedure occurs.2m/s. MODEL OF THE WIND TURBINE DG 80 Induction Generator Wind Turbine ~ To grid SVC(FC-TCR) Figure 8. the wind turbine operating condition can be categorized into three regimes: start-up regime. A real power and reactive power tracking control scheme is presented. trix/vector concept based induction machine model is developed and simulated with Simulink. SVC and general load models are described in the following subsections. a type of static var compensator (SVC).1: Diagram of a wind turbine DG consisting of a wind turbine. The wind turbine induction generator with SVC (called wind turbine DG system) is illustrated in Figure 8.1. sub-rated power regime and rated power regime [88].

R is the wind turbine rotor radius. and the tip speed ratio is: Cp = (0.00184(λ − 3)β (8. SVC consists of a ﬁxed capacitor and a thyristor controlled reactance (FC-TCR).5) (8. A is the blades’ swept area.2 Induction Generator The electrical equations of induction generator model in the arbitrary reference frame can be expressed as: ω p ψ + ψ ωb ds ω b qs ω p vds = rs ids − ψ qs + ψ ds ωb ωb ω − ωr 0 p 0 0 )ψ dr + ψ 0 vqr = rr i0 + ( qr ωb ω b qr p ω − ωr 0 0 0 )ψ qr + ψ 0 vdr = rr i0 − ( dr ωb ω b dr vqs = rs iqs + The mechanical part is expressed in per unit as: p 1 (Te − Tm ) ωr = ωb 2H and the torque equation is expressed as: Te = ψ 0 i0 − ψ 0 i0 qr dr dr qr The detailed induction machine model is given in Appendix A.3) 8.2) is used to compensate the reactive power of the induction generator.3β ΩR Vω ¶ − 0.2) λ= where Ω is the mechanical angular velocity.2.3 Static VAR Compensator (SVC) SVC (see Figure 8. Cp is a power coeﬃcient of the blade which is a function of the blade pitch angle β.6) (8. Vω is wind speed. MODEL OF THE WIND TURBINE DG 81 where.44 − 0. 8.CHAPTER 8.0167β) sin and tip speed ratio λ is: µ π(λ − 3) 15 − 0. (8. ρ is the air density.2.9) (8.7) The ωb is the base electrical angular velocity used to calculate the inductive reactances.4) (8. The reactance of the TCR is a function of the ﬁring angle α[1]. (8.8) (8. .

xL is reactance. The load active power PL and reactive power QL can be expressed as: PL = C1 (Vbus )np QL = C2 (Vbus )nq (8. and Beq is the coeﬃcient of the TCR reactance. and it can be expressed as: xeq = xC π (σ xC xL − sin σ) − xL (8.4 Static and Dynamic Load Models The static load can be represented as a nonlinear function of load bus voltage Vbus . The adjustable coeﬃcient of the reactance is: 0 < Beq < 1 (8.10) where.2: Diagram of the SVC (FC-TCR) structure. 1 xLeq = = 2(π − α) + sin(2α) 1 · π xL σ − sin σ 1 1 = Beq · π xL xL (8.13) (8.14) .11) The SVC equivalent reactance (xeq )is determined by the combination of FC and TCR in parallel.12) 8.2. σ = 2(π − α).CHAPTER 8. MODEL OF THE WIND TURBINE DG V I 82 XC XL Figure 8.

MODEL OF THE WIND TURBINE DG 83 wind-t urbine Cont rol set P* β wind turbine induction generator measuring P grid ~ measuring V.4. As a dynamic load. The control loop of the SVC is shown in Figure 8.CHAPTER 8. the exponent values np = nq = 2. The SVC is used to compensate the reactive power. The wind speed is assumed to be in the rated power regime. Vqr = Vdr = 0. thus stabilizing the commutating bus voltage.5) and the voltage regulator loop (see Figure 8.4) respectively.3 [84].3 Control Unit Model The general structure of a wind turbine induction generator with SVC and the control unit is shown in Figure 8. . The frequency deviation is added as an input to the SVC to improve damping of the system [5]. Q set Q*(V*) SVC Control α SVC Figure 8. The constraints of the output XLeq are adjusted for diﬀerent operating conditions. The voltage regulator control loop is implemented by SVC. The characteristics of diﬀerent loads are given by the exponents np and nq . 8. In this dissertation. The 5th order state equations and the torque equation are the same as the induction generator.3: General structure of a wind turbine DG and its control unit. The ancillary signal is also considered as an input to the SVC to improve oscillation damping of the system. The induction motor is used to represent the dynamic load. the rotor voltage is set to zero. where C1 and C2 are constant. The control unit consists of the blade pitch angle control (see Figure 8. means that the static loads are constant impedances.

5). MODEL OF THE WIND TURBINE DG vs 84 α ref vref + ∑ π + cont roller ka 1+sTa + + ∑ f(α ) π/2 e -sTd 1+sT b xLeq ancillary signal Figure 8. 8. The ancillary signal can be a control signal of the rotor speed deviation of the generators to improve the dynamic stability of the system. . A PI controller is considered to keep the mechanical torque of the wind turbine to be constant (see Figure 8.5: Diagram of the blade pitch angle controller of the wind turbine. the nonlinear induction machine model is used to represent the generator and the dynamic load. A rate limit of ±10 degree/sec is set to limit the pitch angle actuator. The nonlinear model is simulated in Simulink.CHAPTER 8. The eigenvalues of the linearized induction machine are obtained at a certain operating point. The control for the wind turbine DG will be designed in Chapter 9.4: Diagram of the SVC’s voltage regulator control loop. β0 kp ∆pe + + ∑ ∑ ∑ + β + kI 1 s + - ancillary signal Figure 8.4 Induction Machine Simulation In this research.

6. This equation can be further written into the equation (8. i0 . . Assume that the reference frame is arbitrary reference frame and that all quantities are in per unit value.15) is simply expressed as in Figure 8.2 Induction Machine Simulink Model The Simulink model for equation (8.15) 8. x = Ax + Bu ˙ (8.16) 0 8.17) 0 0 0 0 0 rr (8. Where: x = [iqs . MODEL OF THE WIND TURBINE DG 85 The matrix concepts in Matlab are used to simulate the induction machine. The Simulink expressions are also shown below which will use voltages as inputs and currents as the outputs of the model block. ids . i0 .CHAPTER 8. The dynamic models of induction machine is presented below in terms of (8. i0 ]T qr dr 0r −1 1 1 Xss 0 0 XM 0 0 ωb ωb 1 1 0 Xss 0 0 Xss 0 ωb ωb 1 0 0 Xls 0 0 0 ωb B= 1 1 0 0 0 Xrr 0 0 ω Xss ωb b 1 1 0 0 Xss 0 0 Xrr 0 ωb ωb 1 0 0 0 0 0 0 Xlr ωb ω ω −rs 0 0 ωb Xss ω b XM ω ω − ω Xss rs 0 − ω XM 0 b b 0 0 rs 0 0 A = −B · ω−ω r ω−ω r 0 0 0 rr ω b XM ω b XM ω−ωr 0 0 − ω XM 0 0 − ω−ωr Xrr rr ωb b 0 0 0 0 0 The swing equation is: H Te = 2 ωr + TL ˙ ωb The torque equation is: Te = XM (iqs i0 − ids i0 ) qr dr (8.15).4.15). i0s .1 Matrix Concepts Based Induction Machine Model The voltage equations of an induction machine model in the arbitrary reference frame can be written in term of the currents [69].4.

4.7: Swing equation of an induction machine represented in Simulink.16) which is modeled as in Figure 8.6: State equations of an induction machine represented in Simulink. MODEL OF THE WIND TURBINE DG 86 Figure 8.CHAPTER 8.7. the induction machine and the power system network are interconnected and as long as the initial condition is set. use Newton-Raphsom method to calculate the rotor speed ω r . dynamic simulation can be performed. The induction machine serves as a current source to the network. Figure 8.6. It saves not only model building time but also debugging time. The induction machine can be represented as in Figure 8. This model is quite simple and easy to understand. In this model block. then set the steady state variables vector (I). 8.3 Initialization of the Induction Machine The procedure to set the initial condition of induction machine: Given P0 .8. The rotor speed will be fed back to the input of the block in Figure 8. The rotor speed is calculated through equation (8. And the output from the network is the voltage vector. inputs are voltage and rotor speed. output is a current vector. . V as . Thus.

CHAPTER 8.19) ωe ωe 0 Xls )Ias + j XM (Ias + Iar ) ωb ωb Substitute equation (8.18) into (8. given P0 . 0 Xlr 0 rr 2 0 ( s ) +(Xlr )2 = ) Vas R + jX (8. MODEL OF THE WIND TURBINE DG j(ωe/ωb )xls j(ωe/ωb )x'lr I'ar r'r/s + Ias j(ωe/ωb )xM V'ar/s 87 rs + Vas - - Figure 8. Vas Ias 0 rr 0 ( s )2 +(Xlr )2 2 XM 0 rr s 0 rr s j ωe XM ω 0 rr s b 0 +j ωe Xlr +j ωe XM ω ω b b (8. 0 Xlr 0 rr 2 0 ( s ) +(Xlr )2 ∗ Vas P = real(V as I as ) = real(V as = 2 Vas R = f (slip) R2 + X 2 ∗ 2 ∗ Vas Vas ) = real( ) R − jX R − jX (8. 0 Iar = − r0 s j ωe XM ωb 0 + j ωe Xrr ωb r Ias (8.20) ) (8. V as .22) . Therefore.23) (8.8: Equivalent circuit of an induction machine.18) (8. X = Xss + ∗ = R + jX and Ias = R−jX . only slip is unknown and slip_new = slip_old − P0 − f (slip_old) f 0 (slip_old) (8.21) + jXss Vas 2 XM 0 +jXlr = + jXss Vas rs − Vas 0 2 rr 0 XM ( s −jXlr ) 0 rr 2 0 )2 ( s ) +(Xlr 0 2 rr XM s 0 rr 0 ( s )2 +(Xlr )2 + j(Xss + . Var = 0 .25) .19): Vas = (rs + j Ias = Vas (rs + j ωe Xls ) + j ωe XM (1 − ωb ωb = rs − = rs − where R = rs − So. 0 Usually.24) Using Newton-Raphson method.

1 at 2.4 Start-up Procedure and Step Response of Induction Machine To test the induction machine model in Simulink.9: Diagram of a one-machine inﬁnite bus system to test the induction generator.12. rr = 0.10: Rotor speed of an induction machine during the start-up and step response procedure.5 2 2. 8. The initial Tm set point is Tm = 0.1. The parameters of the induction machine: Xls = 0.9).0059.135.6 0.1053(pu).4. the motor start at 0s and the Tm is switched from 0. ´ The start-up procedure and step responses of induction machine are shown in Figures 8.5 1 1. 1.0339.2 0 0. rs = 0.2 1 0.075. the induction generator connected to the inﬁnite bus is simulated (see Figure 8.4 0. Vind = . MODEL OF THE WIND TURBINE DG 88 set Tm ~ induction generator infinite bus Figure 8.161.0s.5.10-8.5 Linearized Induction Machine In this section. The induction machine operates as a generator and the operating point is: Pind = 0.8 rotor speed (pu) 0.5 4 4.CHAPTER 8. 8.2 0 -0.5 5 Figure 8.4. X´ = lr 0. the linearized induction machine model is obtained via the Simulink toolbox. XM = 4.5 3 time (second) 3.1 to -0. H = 0.

.5 4 4. MODEL OF THE WIND TURBINE DG 89 4 3 electrical torque (pu) 2 1 0 -1 -2 0 0.12: State variables of induction machine during the start-up and step response procedure.5 2 2.11: Electrical torque response of an induction machine during the start-up and step response procedure. 8 6 4 state variables (pu) 2 0 -2 -4 -6 -8 0 0.5 3 time (second) 3.5 5 Figure 8.5 2 2.5 5 Figure 8.5 3 time (second) 3.5 1 1.5 4 4.5 1 1.CHAPTER 8.

2 -10. MODEL OF THE WIND TURBINE DG 90 Table 8.13-8. ∗ The initial operating condition: Sbase = 4. ∗ Case I: t=0. Ki =6.3. thus balance the terminal voltage of the induction generator.2.1). Pwind increases from 0.14) show the dynamic responses of the wind turbine DG.4 -32.0516. As a characteristic of the induction machine.476 λ7 -170.02638 and QSV C = 0. it absorbs the reactive power from the system. The two complex conjugate pairs are the “stator” eigenvalues and the “rotor” eigenvalues.0M V A. The eigenvalues are shown in Table 8.40 1. The test system is shown in Figure 8. the induction generator can only generate the real power.0. however the ﬁrst order feedback controller is used to track the voltage through the SVC.548 ± 375.3. Td =0. Tb =0. Sslip = −0. The wind turbine transfers the wind energy to the electric power through the mechanical torque.1053 to 0.16kV . It supplies the controllable reactive power.5.234 λ6 -16.938i λ5 -59.CHAPTER 8. Figures (8. The real power is tracked by the PI controller. The simulation results show the reactive power output decreased since the voltage drop along the distribution line increased.96m. Kp =6.36i λ3. induction generator and SVC (see Figure 8.476. and “rotor” pairs change much[69]. Vbase = 4. The SVC is a very important part of the wind turbine DG.2 and the parameters of the SVC are: XC =0. KR =33.1370 and Q∗ wind remains constant. Q∗ wind = 0.1053.5 Integration of Wind Turbine DG The wind turbine DG consists of wind turbine.5s.225.1: Eigenvalues of the linearized induction machine.1. the frequency of “stator” pairs can remain around ωb . Then initial condition for induction generator and SVC each: Pind = Pwind .0014. meanwhile.1. . 8.606 ± 44.0852.137. ρair =1. For diﬀerent operation condition. Qind = −0. This characteristic can help to identify the eigenvalues. Vrated =9. part of it to the induction generator.00417.3m/s. Eigenvalues λ1. Pwind = 0. A wind turbine DG and inﬁnite bus system is simulated to test the real power (P*) and reactive power (Q*) track performance. NO. TR =0. XL =0. The integration of wind turbine DG is studied in this section. The parameters of wind turbine are: R=69.

∗ Figure 8. .13: Real and reactive power responses of the wind turbine DG to track Pwind (Case I). MODEL OF THE WIND TURBINE DG 91 ∗ Figure 8.CHAPTER 8.14: Terminal voltage response of the wind turbine DG to track Pwind (Case I).

8. Figures (8. In the Chapter 9.0852 to 0.15-8. . thus control the DG’s terminal voltage.17) show the dynamic responses of the wind turbine DG. During the controller design. but cannot reach the actually changed values. The model consists of an induction machine. The SVC is used to compensate the reactive power of the wind turbine DG.5s. The good real/reactive power tracking performance of wind turbine DG will be used to improve the dynamic stability of the system. Pwind remains constant and Q∗ wind increases from 0. a wind turbine and a SVC. the wind turbine DG model will be connected to the IEEE-13 system with existing DGs. A practical Newton-Raphson method is used to determine the initial condition of the induction machine. A matrix/vector concept based simulation of the induction machine is described in detail.117. The responses of the real power and the rotor speed of the induction generator illustrate the interaction between the real power and reactive power tracking procedures.15: Real/reactive power responses of the wind turbine DG to track the Q∗ wind (Case II).6 Summary A wind turbine DG model is presented in this chapter. an additional control signals will be input to the control units. MODEL OF THE WIND TURBINE DG 92 Figure 8. ∗ Case II: t=0. The reactive power response shows the output Q can track the set point.CHAPTER 8.

17: Rotor speed of the wind turbine DG to track the Q∗ wind (Case II). . MODEL OF THE WIND TURBINE DG 93 Figure 8.CHAPTER 8. Figure 8.16: Terminal voltage of the wind turbine DG to track the Q∗ wind (Case II).

The distribution system model. The robust stability of the controllers is evaluated. the controllers of a wind turbine distributed generation (DG) are designed to improve the dynamic stability of the system. A dynamic load represented by induction motor is connected at bus 634.94 Chapter 9 Robust Control of Wind Turbine DG 9. in Figure 9. and a SVC are connected. a diesel engine synchronous generator. . gas turbine model and diesel engine model have been discussed in the previous chapters.2 System Conﬁguration The system studied. The linearized model of the entire system at a certain operating condition is obtained. wind turbine pitch control and the combination of the two. a wind turbine induction generator.1. A wind turbine DG is connected to a power distribution system which has other DGs connected as well. Output feedback coordinated controllers are designed through SVC. consists of IEEE-13 node system to which a gas turbine synchronous generator. 9. µ analysis is used to ﬁnd the margin value of the load parameter.1 Introduction In this chapter. The dynamic loads are brought into consideration by modeling the load uncertainty as structured uncertainty. Dominant eigenvalues are identiﬁed using optimal ranking indices. The existing DGs are gas turbine and diesel engine DGs.

The controller design process is covered in the following sections. In the linearized model. A(n × n) is the system plant matrix. 9.1 Ranking Indices of the Eigenvalues Consider a linear system (A. B. C. The eigenvalues of the linearized open loop system are listed in Table 9. B(n× r) is the control matrix and C(m × n) is the output matrix. ROBUST CONTROL OF WIND TURBINE DG 95 Subst at ion 650 646 645 632 633 634 611 684 671 diesel -engine syn gen (DG2) 692 ind motor (dynamic load) 675 Q 680 652 gas-turbine syn gen (DG1) SVC wind-turbine ind gen (DG3) Figure 9.3. in the linearized distribution system. C.CHAPTER 9.1: Diagram of IEEE-13 node distribution system with multi-DGs to evaluate the wind turbine DG. The output performance index can be described as: . the controller is then designed via eigenvalue analysis. 9. The linearization tools in Simulink are used to get linear model (A. u is the r × 1 control input.1) (9. Based on the linearized model (A. C) expressed as: x = Ax + Bu ˙ y = Cx (9. Hence. B. there are a total of 45 state variables. B. D) directly. only two rotors angles are independent.3 Critical Eigenvalues Identiﬁcation The system includes 46 diﬀerential equations. once the three synchronous generators (2 DGs + Substation) are connected in the distribution system.1.2) where x is the n × 1 linearized state variable vector. D).

661145 λ36 -0.11 -33.23 λ26.8694i λ17 -52.3909i -59. Referring to [32][115].6798 -20.8666i λ28.CHAPTER 9.19824 ± 13. The contribution of each .4 -169.1077i λ14.515i λ10.455469 -0. Then the performance index can be expressed as: J= n n XX i=1 k=1 r XZ i=1 ∞ yi (t)T Qyi (t)dt (9.286084 λ39 -0. Eigenvalues λ3. A way to rank the eigenvalues of the linearized system was proposed in [32][115].65105i λ31. For the sake of the convenience.25 -5. R = V H C T QCV and Θ = i Pr H i=1 Γi Γi .13 λ16 λ18 λ20 λ22. assuming the system (A.950i -108.6496 ± 48.1: Eigenvalues of the Eigenvalues -1181.3) t=0 Tik Θik (9.340 λ43 -259.2 λ5.7052 -32.22992 ± 7.08 ± 3428.6 λ9 λ12.32 -0.0238 ± 69.357 -20.139074 ± 3.124190 ± 5. The diagonal matrix of the eigenvalues is Λ = W AV .473i λ7.808 J= where yi (t) is the output of the system and Q is the weight matrix of the performance index. ROBUST CONTROL OF WIND TURBINE DG 96 Table NO.9048 ± 64.29866 -0. To calculate the ranking indices.016 ± 372.3933 λ19 -31.8 -32. λ1.005198 ± 0.3338 -30.29 -1.370 ± 730.38 λ40 λ42 λ44 9. The right and left eigenvector matrices are V and W . NO. deﬁne η ik = Tik Θik + Tki Θki .27 λ30 λ33 λ35 λ37. B.03003i λ34 -0.913 λ45 -239. control matrix is Γi = W Bi .4759 -220.000 open-loop linearized system.595 -22.4) Rik where Tik = − λ∗ +λk is the explicit solution of : ΛH T + T Λ = −R. the critical modes are the eigenvalues with the largest ranking indices.0614 ± 1.0687 ± 407.7551 λ21 -25.07i -142.08311i -16.5730 -2.22492i -5.193158 λ41 -170.09989i -3.15 -27. The ranking indices are used to evaluate the contribution of the eigenvalues to the performance index.1660 λ24. C) has n distinct eigenvalues.

03003i 0. ROBUST CONTROL OF WIND TURBINE DG 97 Table 9.32 =-0.27 =-3. Its speed deviation is very small and is ignored in the performance index. ωdiesel (t).22992 ± 7.1028 -27.124190 ± 5.03003i. and three outputs are selected to calculate the ranking indices y = (ωgas (t).2 Dominant Eigenvalues Identiﬁcation The rank indices are used to identify the critical eigenvalues of the system. usvc ).124190 ± 5.4 & 8. The two inputs are control signals of wind turbine and SVC (see Figures 8. .29 =-1.5) u = (uwid .15 λ26. The performance index is expressed as: Z ∞ (10ωgas (t)2 + 10ω diesel (t)2 + ωwind (t)2 )dt J= 0 (9.22492i 0.32 Critical eigenvalues of the open linearized system.3. The objective of the dynamic stability control is to improve the oscillation damping when the system is subjected to a ground fault disturbance. The weights between the weights is to balance the inertia and size between the DGs.515i 0. λ7.2: NO.8694i 0.5) The absolute value ρi = |τ (λi )| is deﬁned as the ranking index of λi to the performance index. A is 45×45.65105i 0. and C is 3×45.6496 ± 48.1252 eigenvalue to the performance index can be summarized as: n X 1 η ik τ (λi ) = η ii + 2 k=1. λ28. 1]. ω wind (t)).22492i. In the linearized system of the distribution system with DGs.22992 ± 7. 9.k6=i (9.8 λ14. connecting the distribution line to the transmission system.CHAPTER 9.65105i.139074 ± 3. 10.0687 ± 407.2258 -3. B is 45×2.29 λ31.27 λ28. The dominant eigenvalues related to the oscillation can be selected out: λ26. The performance index is composed of the speed deviations of the DGs. Eigenvalues Ranking Indices(×10−3 ) -32.02227 -0. A large inertia synchronous generator represents the distribution substation. and λ31.6) The output weight matrix is Q = diag[10.139074 ± 3.05226 -1.

the diesel engine DG and the wind turbine DG.1. The objectives of the controller are to adjust the reactive power output of the SVC. (9.8423i.4.1423 ± 2.32 =-1. and to adjust the real power of the wind turbine. For the SVC only input: h i K = −125. and c ª © T (0) .65765i. ROBUST CONTROL OF WIND TURBINE DG 98 9.8) C T K T RKC + Q. see Figure 9.4 9. Deﬁne the performance index [30][72]: Z ∞ J= (xT Qx + uT Ru)dt 0 h i k1 k2 k3 . (9. The above equations are solved iteratively to obtain the gain matrix. The output feedback which will be used to design a controller has the following form: u = −Ky where three types of controllers are designed as: 1) three outputs and only SVC input (SIMO): Ksvc = h i k1 k2 k3 . X = E x(0)x with Q = QT ≥ 0 and R = RT ≥ 0. 0 = SAT + Ac S + X.29 = -0. where Ac = A − BKC. λ28.5123 and the closed-loop dominant eigenvalues are: λ26.CHAPTER 9.7) 2) three outputs and only wind turbine blade pitch input (SIMO): Kwind = 3) three outputs and both SVC and wind turbine blade pitch input (MIMO): k11 k12 k13 .7039 −13.821 −27.2). For the wind turbine pitch controller only input: .83015 ± 5. In this case. λ31.1 Controller Design Multivariable Coordinated Controller A multi/single-input multi-output coordinated controller is to be designed using multiple inputs.1-9.2 Design Results The optimal gains are obtained through the optimal feedback control algorithm. The optimal gain design equations are: 0 = AT M + M Ac + c 9.4. Consider the linear system represented by equations (9.27 =-3.08721i. K= k21 k22 k23 The same output feedback optimal control algorithm as in Chapter 7 is implemented to compute the gain K.78856 ± 8. K = RT B T M SC T (CSC T )−1 . the three outputs are the rotor speed deviations of the gas turbine DG.

Also. The induction motor is connected at bus 634 to represent the dynamic load.3 shows the rotor speed of the diesel engine DG.8428 ± 5.29 = -1. The nonlinear simulation illustrates that the controller through SVC can eﬀectively improve the dynamic stability of the whole system.32 as with the single SVC.29 = -0. For the sake of simpleness and eﬀectiveness.881 −40. T= 2. λ31. Figure 9.4 shows the rotor speed of the wind turbine DG.25s.5 shows controlled susceptance of the SVC.058 −53. the closed-loop eigenvalues with two control inputs has the similar eﬀects on λ28. λ28. λ31.170 −6.27 = -3. Figure 9. the controller with the single SVC is much more eﬀective than the controller with the single wind turbine control in moving the dominant eigenvalues.8237i. ROBUST CONTROL OF WIND TURBINE DG K= h i 99 and the closed-loop dominant eigenvalues are: λ26.0s. Figure 9.29 .1396 ± 3.1242 ± 5.485 −30. For both SVC and wind turbine blade pitch input (MIMO): −392.2532i. The solid lines represent the simulation results with the controller and the dotted lines represent the case without controller.2 shows the rotor speed of the gas turbine DG.1). The following ﬁgures show the simulation results.0302i.32 = -1.7962i Comparing the three types of controllers and their closed-loop dominant eigenvalues.6510i.413 2.CHAPTER 9. .1476 ± 2.084 K= −50.603 9.2248i. −125.5 Nonlinear Simulation to Test Controller Nonlinear simulations are conducted to verify the controller Ksvc . λ28. Figure 9. a ground fault occurs at bus 632 and is cleared after 0.824 66. λ31. Two cases are simulated: one is with controller. The wind turbine induction generator is connected to the IEEE 13-node distribution system with existing DGs: the gas turbine synchronous generator and the diesel engine synchronous generator (see Figure 9.27 = -5.4015 ± 8. the other is without controller.2303 ± 7.758 The closed loop dominant eigenvalues are: λ26. the feedback controller Ksvc is the best choice to improve the dynamic stability of the system.32 =-0.

3: Rotor speed of the diesel engine DG with/without a controller of the SVC.2: Rotor speed of the gas turbine DG with/without a controller of the SVC. . Figure 9. ROBUST CONTROL OF WIND TURBINE DG 100 Figure 9.CHAPTER 9.

.CHAPTER 9.4: Rotor speed of the wind turbine DG with/without a controller of the SVC.5: Diagram of the susceptance of SVC with controller. Figure 9. ROBUST CONTROL OF WIND TURBINE DG 101 Figure 9.

such as rotor resistance (rr ). M12 . An interconnected system is shown in Figure 9. (9.CHAPTER 9.6 Controller Robustness Analysis The parameters. 9.6.9) (9. stator resistance (rs ) and mutual reactance (XM ). as r2 = ∆v2 .10) . The linear system plant M can be fractionalized into M11 . and M22 . ∆)r1 The notation FL indicates that the lower loop of M is closed with ∆.6. The relationship between r1 and v1 is: v1 = (M11 + M12 ∆(I − M22 ∆)−1 M21 )r1 = FL (M.11) (9. To accommodate the variation of the parameters. Suppose: v1 = M11 r1 + M12 r2 v2 = M21 r1 + M22 r2 and a matrix ∆ relates v2 to r2 .1 Representing Uncertainty Linear Fractional Transformations (LFTs) are used as a tool to represent uncertainty in matrices and systems. the robustness of the controller is evaluated versus parameter uncertainty.6: Diagram of the interconnected system. M21 . This uncertainty is a structural uncertainty and can be represented by the linear fractional transformation (LFT). 9. of a dynamic load vary when operating conditions change. ROBUST CONTROL OF WIND TURBINE DG 102 r1 r2 M11 M21 M12 M22 v1 v2 ∆ Figure 9.

CHAPTER 9. δ) A B E 0 where G = C D F 0 0 The system of Figure 9. the µ of M (jω) is calculated with respect to the uncertainty set δ. A = FL (G. F ∈ Rr×n This is a linear fractional transformation (see Figure 9.6. The structure singular value µ of linear system M with respect to the uncertainty set δ is used to analyze robust stability of the closed-loop system. A0 is a coeﬃcient matrix.13) A is the coeﬃcient matrix. The state space model of the interconnected system is .7).7: Diagram of the linear fraction transformation. where E ∈ Rn×r . where M is a known linear system and δ is a structured perturbation.7 can be recast into a feedback loop system as Figure 9.12) (9. At each frequency. ROBUST CONTROL OF WIND TURBINE DG 103 K + u + B + + 1 s x C y z [A E F 0] w δ Figure 9.2 Parametric Uncertainty For a linear system: x = Ax + Bu ˙ y = Cx (9.8. 9. Let A0 be factored as A0 = EF . Suppose there is an uncertainty of A matrix and it can be written as: A = A0 + A0 δ where δ varies within a range.

8: Digram of the feedback loop.638 0 0 0 0 0 0 −58. giving upper and lower bound functions of frequency. µ of the matrix M (jω) is calculated and the bounds for µ(M (jω)) are computed. y = G u w z x ˙ x 9. ROBUST CONTROL OF WIND TURBINE DG 104 ∆ M Figure 9.323 −17.142 0 0 −40.638 −42.775 16.775 0 −40.6.994 40.142 0 16.CHAPTER 9. If the varying region of δ is within 1. [137].666 0 0 42.3 Analyzing Robust Stability The structured singular value µ of M with respect to the uncertainty set δ is used to analyze robust stability of the closed-loop system. then the system is robust stable when µ < 1. then the perturbed system is stable where σ is the upper bound of δ(jω). The bounds for µ∆ (M (jω)) can be computed by MATLAB µ toolbox. this means that for all perturbation matrices δ with the appropriate structure and satisfying max(σ(δ(jω)) < 1 β.005 change. µ analysis has the following steps: 1. Let rs _mot has a 0.666 0 −17.539 .994 0 0 0 0 0 0 0 A1 = 0 −41. If the peak of µ∆ (M (jω)) is β.323 0 41. Cast the problem into the feedback loop diagram. At each frequency. A has a change of 026×26 026×6 026×13 A − A0 = 06×26 A1 06×13 013×26 013×6 013×13 0 40.

δ1 0 δ2 0 0 That means rs _mot ’s uncertainty δ can be represented as: 0 0 2.142 0 0 −40. Describe the structure of the perturbation of δ 0 δ3 4.where E = [zeros(26. 3). The Matlab commands are shown in Appendix D. 3. 3)].539 0 −1.0378 0 0 0 1 Then A = A0 + EδF. 13)]. . where 1 0 0 0 1 0 0 40. ROBUST CONTROL OF WIND TURBINE DG 105 0.775 0 0 0 0 E1 = .994 40. A1 can be factorized as A1 = E1 × F1 .0378 0 0 0 0 0 0 0 −58. F 1.015 M11 0.CHAPTER 9. E1. F = [zeros(3.775 16.005 0 -2 10 10 -1 10 Frequency (rad/sec) 0 10 1 10 2 Figure 9.02 0.01 0.994 0 −1. Run the Matlab command mu on frequency response to obtain the upper and lower bound of µ.025 0.142 0 16. zeros(3. 26). Calculate a frequency response of M11 .9: Frequency response of M11 . F1 = −40. zeros(13.

9. rr .16.25 0. The frequency response of M11 is shown as in Figure 9.005 increase. Figure 9. Figures 9. XM ) of dynamic loads increase 0.CHAPTER 9.25s. In Figures 9.16 illustrate the simulation results with/without controller. Figure 9.7 Simulation Results of Robustness Analysis To verify the analysis results of the controller (Ksvc ) robustness. rr . So the system is unstable.005 increase.0s. XM ) of the dynamic loads all increase 0.02pu without feedback controller Ksvc .02 increase. . as shown in Figure 9. With the value of rs _mot having a 0.10: Diagram of Mu (µ) as rs_mot has a 0. a ground fault occurs at bus 632 and is cleared after 0.15 mu(M) 0.12 shows the µ upper bound has exceeded the 1. The third case is that the three parameters (rs . the system is still stable.0. the solid lines represent the simulation results with the controller and the dotted lines represent the results without controller.02(pu).9. Figure 9.05 0 -2 10 10 -1 10 Frequency (rad/sec) 0 10 1 10 2 Figure 9. the nonlinear simulation is conducted with the changed parameters of the dynamic loads.1 0. T= 2. The upper bound never exceeds 1 which means when the rs _mot has a 0.2 0. ROBUST CONTROL OF WIND TURBINE DG 106 0. When the parameters (rs .10 shows the plot of the upper bound of singular value µ. the system is still stable since the upper bound of singular value µ doesn’t exceed 1.16 shows the susceptance of the SVC with the controller.11.13-9.13-9.

7 0. 1.5 0.2 mu upper bound 1 0.6 0.2 mu lower bound 10 -1 0 -2 10 10 0 10 1 10 2 Figure 9.6 mu(M) 0.4 0.2 0.02pu.1 0 -2 10 -1 0 1 2 10 10 Frequency (rad/sec) 10 10 Figure 9.3 0.12: µ upper bound and lower bound of system with the three parameters (rs .4 0.11: Diagram of Mu (µ) as rs_mot has a 0.9 0.02 increase.8 0. rr . XM ) of dynamic loads increasing 0.4 1. (X axis: frequency (rad/sec)) .8 0.CHAPTER 9. ROBUST CONTROL OF WIND TURBINE DG 107 1 0.

Output feedback optimal control is implemented to assign the critical eigenvalues. The wind turbine DG model is connected to the IEEE-13 node distribution system together with existing gas turbine and diesel engine DGs. When the controller is added to the system. The SVC and the blade pitch are considered as the control inputs. 9.13: Rotor speed of the wind turbine DG with modiﬁed parameters of dynamic loads when a fault occurs at bus 632 (dotted line: without controller. The critical eigenvalues are identiﬁed via the ranking indices technique. ROBUST CONTROL OF WIND TURBINE DG 108 9 8 7 6 Rotor Speed (pu) 5 4 3 2 1 0 0 1 2 3 4 5 6 time (second) 7 8 9 10 Figure 9. the rotor speeds of the gas turbine and diesel engine DGs show that the controller can improve the oscillation damping of the entire system as well.13. In Figure 9.15. The design results show that the SVC controller is better than the blade pitch due to SVC’s rapid response characteristic.14 and 9. solid line: with controller). In Figures. The resulting controller’s . 9. A linearized model is obtained for the speciﬁc operation condition.CHAPTER 9.8 Summary A robust controller is designed for the wind turbine DG to improve the oscillation damping of the system. the rotor speed of the wind turbine DG shows that the system loses the stability without the controller when the dynamic loads’ parameters are changed. The µ analysis and the nonlinear simulations show that the coordinated controller is robust in improving the dynamic stability of the system over a relatively large parameter range in the dynamic loads. the system remains stable.

solid line: with controller).015 1.005 1 0.99 0 1 2 3 4 5 6 time (second) 7 8 9 10 Figure 9.025 1.995 0. solid line: with controller).995 0.99 0. 1.01 Rotor Speed (pu) 1.CHAPTER 9. .14: Rotor speed of the gas turbine DG with modiﬁed parameters of dynamic loads when a fault occurs at bus 632 (dotted line: without controller.98 0.015 1.02 1.01 Rotor Speed (pu) 1.975 0 1 2 3 4 5 6 time (second) 7 8 9 10 Figure 9.15: Rotor speed of the diesel engine DG with modiﬁed parameters of dynamic loads when a fault occurs at bus 632 (dotted line: without controller.985 0.005 1 0. ROBUST CONTROL OF WIND TURBINE DG 109 1.

5 1 Susceptance of SVC (pu) 0. The nonlinear simulation results illustrate the controller via the SVC can improve the dynamic stability of the system.5 0 1 2 3 4 5 6 time (second) 7 8 9 10 Figure 9.5 -1 -1. . robustness is evaluated over a relatively large range of parameter uncertainty in the dynamic loads.5 0 -0. ROBUST CONTROL OF WIND TURBINE DG 110 1.16: Susceptance of SVC with a controller when a fault occurs at bus 632.CHAPTER 9.

1. Usually. As an example to . A centralized dynamic stability controller will be used to improve the dynamic stability The coordinated control through the rapid fuel cell inverter and the SVC of wind turbine DG are studied.2 Interaction of DGs A distribution system with multi-type DGs is shown in Figure 10. more and more small distributed power generation units will be connected to the grid. fuel cell power plant. is added to the system. diesel engine. a new DG. in power distribution systems the distances between the ends of feeders and the substation is relatively short and it is practical to implement a centralized control for the entire system. Compared to the studied system in Chapter 9. The controller uses the multi-variable input to assign the eigenvalues. The dynamic responses of the distribution system with diﬀerent real power penetration of DGs are investigated to evaluate the coordinated controller. In this system. 10.1 shows a power distribution system with multiple types of distributed generations (DGs): gas turbine. The interaction between DGs and the impacts of DGs on the dynamic stability of the grid will be evaluated. all the four DG models which were developed in the previous chapters are connected to the system together. wind turbine and fuel cell DGs.111 Chapter 10 Coordinated Operation of Multiple Types of DGs 10.1 Introduction In the future. Figure 10.

The eigenvalues of the linearized model are listed in Table 10. To evaluate the interaction of the DGs.16kV .1969 + j0.3 second. with appropriate location can eﬀectively improve the dynamic stability . evaluate the interaction between DGs. the dotted lines represent the case without SOFC and the solid lines represent the case with SOFC.4 demonstrate the impact of the newly connected SOFC power plant on the dynamic stability of the system with existing DGs. Sdiesel = 0.0852.0098. Swind = 0. In Figures 10.2-10.1: Diagram of IEEE-13 node power distribution system with multi-type of DGs to investigate the coordinated operation.0666.2-10.29 . The cases were done under the condition that a three-phase ground fault occurs at bus 675 and is cleared after 0.4.CHAPTER 10. the impact of the solid oxide fuel cell (SOFC) power plant on the linearized model and nonlinear dynamics is investigated in this section. Comparing the eigenvalues in the Tables 9.27 . The initial conditions are: Sbase = 4.0M V A.1. λ31. COORDINATED OPERATION OF MULTIPLE TYPES OF DGS 112 Subst at io n 650 646 645 632 633 634 611 684 671 diesel -engine syn gen (DG2) 692 ind motor (dynamic load) 675 Inverter Q 680 652 SOFC SVC wind-turbine ind gen (DG3) gas-turbine syn gen (DG1) Figure 10. the fuel cell can improve the damping of the critical eigenvalues: λ26.0656 + j0.0125. λ28. such as fuel cell.32 . diesel engine and wind turbine DGs remain same as before (see Figure 9.1054 + j0. The interaction between the SOFC power plant and other DGs illustrate the newly connected DGs.1031 + j0. This linearized system is an open-loop system. Vbase = 4.1 and 10. Ssof c = 0. Sgas = 0. the initial conditions and the parameters of the gas turbine. Figures 10. Dynamic simulation is also done to evaluate the interaction between DGs.1.1).

COORDINATED OPERATION OF MULTIPLE TYPES OF DGS 113 Table 10.03831 -0.3336 -30.000 -292.2872 ± 5.01277 -0.3094 -0.339 -0.004056 ± 0.1741 ± 3. λ3.8649i -1.4958 -4061.9442i -5.95i -141.29 λ31.3014 -0.340 -259.2 λ5. λ1.752 ± 373.022i -32.90 ± 726.8 λ10.3811 ± 6.38 λ40 λ42 λ44 λ46 λ48 λ50 λ52 Eigenvalues -1183.8021i -0.19316 -170.913 -249.1225 -0.4 λ7.32 λ34 λ36 λ39 λ41 λ43 λ45 λ47 λ49 λ51 Eigenvalues -173.476i -32.9056 ± 64.5015i -52. NO.5702 -31.25 λ28.4546 Figure 10.6929 -20.9941 ± 69.13 λ16 λ18 λ20 λ22.64663 -0.0993i -3.0614 ± 1.79003 -25.506 -0.11 λ14.1382i -1.23 λ26.2469i -27.6960 -32. .1: Eigenvalues of the linearized coordinated system.2028 ± 407.15 λ17 λ19 λ21 λ24.2437 -2.357 -20.08155i -16.6 λ9 λ12.CHAPTER 10.2006 ± 13.28 ± 3428.3907i -59.315i -108.736 NO.2: Rotor speed of the gas turbine DG with/without SOFC connected.8289 ± 49.27 λ30 λ33 λ35 λ37.4924 -5.595 -22.3436 -0.4759 -220.

.CHAPTER 10.3: Rotor speed of the diesel engine DG with/without SOFC connected. COORDINATED OPERATION OF MULTIPLE TYPES OF DGS 114 Figure 10. Figure 10.4: Rotor speed of the wind turbine DG with/without SOFC connected.

Here. The the gains are calculated by the optimal algorithm: h i Ksvc = −212.8 λ10.9941 ± 69. COORDINATED OPERATION OF MULTIPLE TYPES OF DGS 115 Table 10. Ksvc = k1 k2 k3 . ∆ω diesel .16772 -3. A coordinated control is then designed with the output feedback optimal algorithm.27 λ31.1) 0 The critical eigenvalues and their ranking indices are listed in Table 10. Then the controller with only SVC is: usvc = −Ksvc × y.1382i 0.584 −89.022i 0.5015i 0.2. h i and the controller with only fuel cell is: uf c = −Kf c × y.8289 ± 49.702 −30.3 Critical Eigenvalues and Controller Design The critical eigenvalues are identiﬁed with the optimal algorithm. The deﬁnition and optimal algorithm of critical eigenvalues has been presented in the previous chapter. Kf c = k1 k2 k3 . and the real power set point of the fuel cell.CHAPTER 10.03633 -0.2469i 0. .32 Eigenvalues of the open linearized system with multi-DGs. the optimal gains for the SVC and fuel cell are deh i signed separately. h i Kf c = −302.02165 -32. Eigenvalues Ranking Indices(×10−3 ) -32. The performance index of the critical eigenvalues is expressed as: J= Z ∞ (10ωgas (t)2 + 10ω diesel (t)2 + ωwind (t)2 )dt (10.2: Critical NO. The control inputs are reactive power set point of wind turbine (SVC). 10.4423 . Then the gain matrix is: usvc = −Kcoor × y uf c k11 k12 k13 . λ7.061 34. ∆ω wind ].3811 ± 6.02022 -27.9442i 0.844 −31. the outputs are the rotor speed deviations of the gas turbine.11 λ14. Kcoor = k21 k22 k23 Before designing the optimal gain Kcoor .324 . diesel engine and wind turbine DGs: y = [∆ωgas .15 λ26.2028 ± 407.07266 of the whole system.1741 ± 3.

893i -36. −98.062i Figure 10.022i -54.8 -32.775i λ10. NO.247i -33.332 ± 398. .915 −41. The dynamic simulation results with/without coordinated controller are shown in Figures −49.3: Critical Eigenvalues of the multi-DGs system with diﬀerent controllers.096i λ31.502i -22.32 -0. Open loop with Kcoor with Ksvc with Kf c λ7.3.893i λ26. Kcoor = .056 ± 417.940i -4.496 −31.5: Rotor speeds of the gas turbine DG with diﬀerent controllers.CHAPTER 10.994 ± 69.390i -4.307 (10.5-10.15 -27.588 ± 7. The results show that the coordinated controller is more eﬀective to improve the dynamic stability of the system with multi-DGs.944i -4.203 ± 407.11 -32.315i -37.209 ± 3.138i -3.538 ± 71.7).634 ± 52.973 ± 7.829 ± 49.420i -33. The rotor speeds of gas turbine with only SVC or fuel cell are compared in Figure 10.286i λ14.147 28.942 ± 68.810 4.27 -3.461 The critical eigenvalues with three types of controllers are listed in Table 10.874 ± 70.381 ± 6.210i -30.943 ± 64.927i -3.092i -1.973i -1.903 ± 64.5.371i -22.820 ± 409. COORDINATED OPERATION OF MULTIPLE TYPES OF DGS 116 Table 10.175 ± 1.887 ± 7.174 ± 3.630 ± 3.

6: Rotor speed of diesel engine DG with/without coordinated controller.7: Rotor speed of the wind turbine DG with/without coordinated controller. COORDINATED OPERATION OF MULTIPLE TYPES OF DGS 117 Figure 10. . Figure 10.CHAPTER 10.

224i.264 −24.0852i Sf c 0.130i -3.0125i power penetration case 2 0.745 ± 50.1152i 0.5: Critical eigenvalues of the open linearized systems with diﬀerent real power penetration.3365i Sgas 0.1477 ± 69.0852i 0.914 ± 50.1382i case 2 -32.2625 + 0.6).1000 + 0.5015i -3.32 case 1 -32. λ14.8 =-37. case 1 Ssub 0.002i -0.0098i Sdiesel 0.013 .1000 + 0. COORDINATED OPERATION OF MULTIPLE TYPES OF DGS 118 Table 10. Three cases are evaluated (see Table 10.4 Impacts of Diﬀerent Real Power Penetrations To deepen the investigation of the coordinated operation of the multi-DGs.289 ± 7.079i -3.1000 . loop system eigenvalues of the case 2 are: .1054 + 0.030 10.909i -33.0666i Swind 0.400 ± 70. NO.756 ± 6.1603 ± 3.102i.3937 + 0.0852i 0. To investigate the robustness of the feedback controller. But the controller is not suitable for the heavier penetrations.1054 + 0.1054 + 0.15 λ26.486i -0.0656 + 0. Case 1 has the normal real power penetration and the controller Ksvc is designed under this condition.9442i -0. The new critical closedλ7.118 ± 2.2469i -27. The feedback controller of the case 2 with heavy real power penetration is redesigned.987i.4865 + 0.974 ± 48.062i -32. h i The gain of the case 2 is: Kf c_case2 = −92.3406i -27. the impacts of the diﬀerent real power penetrations of each DG on the dynamic stability are studied in this section.5.074i -27. case 3 0.0467 + 0. In case 2.3811 ± 6.2063 + 0. the real power penetration of each DG is heavy and almost all the loads are supplied by the four DGs. the The closed loop eigenvalues show that the designed controller is suitable for the lighter real power penetration of DGs. λ26.15 =-4.1969 + 0.11 =-3.8289 ± 49.206 ± 407.8 λ10.0304i 0.27 = -3.+ 0.2028 ± 407.214 ± 410.05i of the multi-DGs.3061i 0.2844i 0.1309 ± 406. the closed-loop system eigenvalues of diﬀerent real-power penetration are calculated (see Table 10. The DGs’ real power penetrations of Case 3 is near Case 1.2912i 0.CHAPTER 10.0.4). In Case 2.11 λ14.948 ± 69.27 λ31.1031. The open-loop system eigenvalues of the three cases are listed in Table 10.4388 + 0.022i -32.4: Diﬀerent real NO. λ7. λ10.9941 ± 69.0265i 0.322 ± 6.260i.1741 ± 3.749i case 3 -32.0750i Table 10.032i 10.

829 ± 49.973 ± 7.183i λ31.8 -32.361i λ14. The study shows that the appropriate new DG location can improve the dynamic stability of the whole system.146i -3.994 ± 69.27 -3.314i λ26.859 ± 7.892i The closed-loop eigenvalues of the redesigned controller are close to eigenvalues of cases in Table 10.6: Critical Eigenvalues of the multi-DGs system with diﬀerent controllers. The output feedback controller is robust with respect to the diﬀerent real power penetration of each DG.166i -33.775i -37.658 ± 2.116 ± 51.538i -36.896i -1.502i -4.062i -0.32 -0.138i -1.174 ± 3.992 ± 70.381 ± 6.11 -32.096i -3. . COORDINATED OPERATION OF MULTIPLE TYPES OF DGS 119 Table 10.421 ± 3. The controller design procedure illustrates that the optimal feedback control is robust with respect to diﬀerent real power penetration of the DGs. 10.CHAPTER 10. Open loop case 1 case 2 case 3 λ7.799 ± 6.818 ± 70.060i λ31.874 ± 70.558i λ10.5 Summary The coordinated operation of the distribution system with multi-DGs is studied in this chapter.249 ± 53.639 ± 410.6.15 -27. The coordinated controller is designed with the optimal output feedback algorithm.426 ± 409.203 ± 407. NO.944i -3.634 ± 52.663 ± 2.630 ± 3.247i -33.893i -3. The closed-loop eigenvalues show controller Ksvc can improve the dynamic stability for the operating point with heavy real power penetration of DGs.286i -33.32 =-0.820 ± 409.022i -36.115i -5.

the three-phase distribution system with DGs was studied using the Power System Blockset (PSB). Thyristor controlled braking resistor (TCBR) is used to improve the damping of the distribution system with DGS. the gas turbine. In power distribution system. 4.1 Introduction One challenge of the distributed generations (DGs) research is the three-phase unbalanced characteristics. 3. 2. dynamic stability control is raised as a new topic while the DGs are widely used. The IEEE-13 node system is simulated in detail. In this chapter. to implement the variable structure control(VSC) to control the TCBR to improve the dynamic stability. . The objectives of this chapter are: 1. The three-phase dynamic simulation in PSB is done to obtain the results. diesel engine and excitation system are connected to the full-order synchronous generator in PSB. to simulate the three-phase TCBR and IEEE-13 node feeder in PSB. The three-phase load ﬂows with/without DGs are solved. In this chapter. to combine the models with the PSB’s synchronous generator. to solve the three-phase power ﬂow to set the initial condition of each electric component.120 Chapter 11 Three-Phase Distribution System Study 11.

1: Diagram of the excitation system. The TCBR model and its impedance characteristics are also studied.1 shows the excitation system model. The local control input signals are frequency deviation and voltage. Figure 11.2 System Models The excitation system of the gas turbine and diesel engine DGs is presented in this chapter. The dynamic process during turning on the thyristor for each phase can be expressed as in Figure 11. the ﬁeld voltage of the synchronous machine is controlled by the exciter output “Vf ”. 11. 11. In Figure 11. During the logic turn on process. and ZΣ is the equivalent impedance of the distribution system. equivalent voltage e keeps constant during each turn on process.1 Excitation System Model The same excitation system model is used for both the gas turbine DG and the diesel engine DG.2. THREE-PHASE DISTRIBUTION SYSTEM STUDY Vref Transient Gain Reduction VT + 1+sTC 1+sTB KA 1+sTA VRmax Vf 121 VRmin Figure 11.2.CHAPTER 11.1. .2 Thyristor Controlled Braking Resistor (TCBR) TCBR is a shunt controllable resistor. 11.2.2. In the system computation. In Figure 11. the voltage sag is very large without the reactance x0 . the input signal is the terminal voltage of the synchronous generator and the output is the exciter voltage Vf .

**CHAPTER 11. THREE-PHASE DISTRIBUTION SYSTEM STUDY
**

Z∑=R∑+jX∑ V

122

e

α i α

r0

x0

Figure 11.2: Equivalent circuit of one phase TCBR. The average impedance of each phase is [128]: γ − sin γ = = where γ = 2(π − α + θ) is the conducting angle; α is the ﬁring angle; θ is delay angle due to the reactance; z0 = r0 + jx0 is the ﬁxed impedance of the TCBR; z = r + jx is the actual braking impedance, and is controlled by the ﬁring angle α ; To avoid switching failure of the thyristor, the constraint on the ﬁring angle is: 0 < α < 150◦ Thus the actual braking impedance z of the TCBR is limited as: z0 < z < ∞ (11.4) (11.3) πz0 z π(r0 + jx0 ) r + jx (11.1) (11.2)

CHAPTER 11. THREE-PHASE DISTRIBUTION SYSTEM STUDY

123

11.3

Controller Design

A variable structure controller is designed based on the analysis of the distribution system with multi-machine. In this multi-machine, the reference machine, substation, is very large.

11.3.1

Basic Concepts of Variable Structure Controller

In general, a nonlinear power system model can be expressed as[128]: ˙ X = A(X) + B(X)U (11.5)

where X=(X1 , X2 , ..., Xn−1 , Xn )T is the the state variable vector. A(X) and B(X) are the nonlinear vector function. To design a VSC based controller, the state space is divided into two parts by the linear switching hyperplane. The hyperplane is deﬁned as: S = CT X = 0 where C is a constant vector. Then, (11.6)

˙ ˙ S = CT X = C T A(X) + C T B(X)U

(11.7)

The reaching model or reaching phase is that the state will move toward and reach the sliding surface. To get asymptote system, a reaching condition [40] should be satisﬁed: ˙ SS < 0 (11.8)

11.3.2

Dynamic Braking Control Strategy

In the multi-machine system, the swing equation can be expressed as: Mi p2 δ i + Di pδ i = Pmi − Pei (t) − Pbi (t) (11.9)

where Mi , Di , Pmi , Pei , Pbi , are the inertia and damping constants, input power, electrical output power and the power absorbed by braking resistor. Neglect the damping term and select one machine (substation) as the reference machine, the swing equation becomes:

CHAPTER 11. THREE-PHASE DISTRIBUTION SYSTEM STUDY

124

p2 δ ir =

**1 1 (Pmi − Pei (t)) − (Pmr − Per (t)) Mi Mr 1 1 − Pbi (t) + Pbr (t) Mi Mr
**

1 Mr

(11.10)

where δ ir =δ i − δ r . Since the inertia of the substation is very big, 11.10 can be neglected.

= 0, the reference terms in

To improve the damping of the oscillation of the system, the performance index is chosen: Z J = min( system is expressed as: 1(on) u(t) = 0(of f ) if Σb > 0 if Σb < 0

tf

dt)

(11.11)

t0

From the conclusion in [97] [98], the optimal control strategy of TCBR in multi-machine

(11.12)

where Σb is the control area. Since the substation can be considered as an inﬁnite bus, the frequency deviation of substation can be ignored. The control can be simpliﬁed as: 1(on) u(t) = 0(of f ) if ∆ωi > 0 if ∆ωi < 0

(11.13)

11.3.3

Three-Phase Control Strategy

The three-phase one-line diagram of TCBR and its control structure is shown in Figure 11.3 [81]. The wye-connected TCBR is controlled on or oﬀ by the control unit. The net available energy for acceleration of the DG during a disturbance is absorbed by the TCBR. In Figure 11.3, the “limits unit” sets the varying range of ﬁring angle α to avoid switching failure of the thyristor. The output of “block unit” will be “1”, either when ∆ω is negative or when |∆ω| is very small. A three-phase TCBR has diﬀerent capacities to absorb the real power depending on the ﬁring angle of each thyristor. The series connected x0 is a small size reactance to prevent the large current while the thyristor is turned on.

In the test.4 Test System of TCBR The IEEE 13 node test feeder [24] is used as the test system to investigate the dynamic characteristics of the distribution system with two DGs and the eﬀectiveness of the TCBR on the stability of distribution system. Table II shows the three-phase voltages and currents of each line with the DGs connected. the voltage and current of “Line 650-632” are measured at bus 632. when no DGs are connected to the distribution system. For example.CHAPTER 11.4 shows the test system with the two DGs. In Figure 11. c pulse-1 + 1500 limits unit α 6-pulse generator 6-pulse (to six thyristors) Va. thus the substation transformer is 138kv/4. the voltage (V) and current (I) are measured at the second end of each line.16kv. b. THREE-PHASE DISTRIBUTION SYSTEM STUDY Gas turbine or Diesel engine Synchronous generator 125 A phase B phase C phase ~ measure ∆ω measure Va. the connected transmission system is simulated as 138kV system. Three-phase power ﬂow is computed before the dynamic simulation and the three-phase voltages and currents of each line are shown in Table I. DG1 is a gas turbine generator and connected at bus 671. In Table I and II.c r0 r0 r0 pulse-3 pulse-5 pulse-4 pulse-6 pulse-2 Kω block unit - block signal x0 x0 x0 Figure 11. 11. Three-phase power ﬂow is studied ﬁrst. One TCBR is installed at the terminal of DG1. .b.4. the two DGs are shown connected to the distribution system. Figure 11. DG2 is a diesel engine generator and connected to bus 633.3: Diagram of the three-phase TCBR and its control unit.

5∠0.40◦ V-Phase B 2515∠-118.21◦ 0 192.02◦ 53.45◦ 64.23∠-0.60◦ 2547∠-117.3∠-84.85◦ I-Phase C 444.6∠-112.2∠-49.8∠-33.1∠121.24◦ 2327∠120.30◦ 70.38◦ 116.99◦ 170∠115.3∠118.33◦ 65.21◦ 192.98◦ 0 0 2409∠-1.8∠84.85◦ .16∠60.3 ∠-174.CHAPTER 11.16◦ 2316∠120.50◦ 154.79◦ 0 0 0 2550∠-116.7∠91.65◦ 233.79◦ 68.69∠-53.86◦ 270.52◦ 0 0 0283∠-31.19∠158.52◦ 2516∠-118.68◦ 545.35◦ 2406∠120.21◦ 2504∠-118.27◦ 2316∠120.87◦ 136.4∠-117.98◦ 682.7∠-51.59◦ 68.67◦ 2407∠-1.4∠-139.20◦ 2409∠-1. Line 650-632 632-633 633-634 632-645 645-646 632-671 671-692 692-675 671-684 684-611 684-652 DG1 DG2 V-Phase A 2458∠0.6∠29.6∠127.8∠127.22◦ 201.32◦ 2407∠120.20◦ 2400∠-1.30◦ 0 0 0 192.31◦ 70.83◦ 2395∠121.15◦ I-Phase B 306.69◦ 149.27◦ 2327∠120.87◦ 136.6∠150.13◦ 36.82◦ 2512∠-118.70◦ I-Phase A 403.69∠-53.6∠-154.1∠0.63◦ 224.00◦ 279.70◦ 0 63.97∠-35.27∠61. Table 11. THREE-PHASE DISTRIBUTION SYSTEM STUDY 126 Subst at ion 650 646 645 632 633 634 DG2 611 684 671 692 675 DG1 680 652 α α r0 x0 Figure 11.01◦ 0 2338∠121.7∠68.65◦ 2445∠0.44◦ 2370∠121.76∠127.06◦ 521.4: Diagram of IEEE-13 node distribution system with multi-DG and TCBR.14◦ 267.77◦ 2547∠-117.77◦ 2478∠-117.23◦ 2321∠120.1: Three-phase current and voltage of distribution system with DG.56∠-119.31◦ 2423∠-0.13◦ 136.6∠7.62◦ V-Phase C 2405∠120.7∠-171.77◦ 2558∠-117.68◦ 65.20◦ 2425∠1.6∠70.29◦ 0 2393∠-1.7∠-13.00◦ 154.54◦ 162.

and initial real power size of DG2 is 0.6 0.6 show the dynamic simulation results of the terminal voltage and stator speed deviation for DG 1 and Figures 11.6 Summary In this chapter.5-11.2 0 0 1 2 3 4 5 6 7 8 time (s) Figure 11. At t =0.1s. Figures 11.7-11.671pu.7 show the voltage of DG1 and DG2. unbalanced three-phase dynamic simulation of a distribution system with two DGs connected is performed using MATLAB/Simulink.5: Terminal voltage of DG1 as a fault occurs at bus 684 with/without TCBR. 11.3pu. This is because the fault location is close to the DG1.8 show the similar results for DG2. thus the TCBR can improve the dynamic characteristic of the whole distribution system. THREE-PHASE DISTRIBUTION SYSTEM STUDY 127 1. Nonlinear models for gas turbine and diesel engine driven synchronous generators are implemented to study the DGs’ impact on the power .CHAPTER 11. the fault occurs at bus 684 and the fault is cleared at t=0. The simulation results show that one TCBR close DG1 can eﬀectively damp the oscillation of both DG1 and DG2. and the initial real power size of DG 1 is 0.8 with TCBR 0.5 and 11.4 1. The magnitude of the voltage minimum for DG1 is lower than for DG2. Figures 11.4 0.5s.5 Dynamic Simulation Results The two DGs are connected at buses 671 and 633. 11.2 no TCBR 1 voltage (pu) 0.

2 no TCBR 1 voltage (pu) with TCBR 0.4 1.4 0.6 0.7: Terminal voltage of DG2 as a fault occurs at bus 684 with/without TCBR.2 no TCBR 1 frequency (Hz) 0. 1.4 0.2 0 0 1 2 3 time (s) 4 5 6 7 8 Figure 11.8 0. THREE-PHASE DISTRIBUTION SYSTEM STUDY 128 1.6 0.CHAPTER 11. .2 0 0 1 2 3 4 5 6 7 8 time (s) Figure 11.4 1.6: Rotor speed deviation of DG1 as a fault occurs at bus 684 with/without TCBR.8 with TCBR 0.

8: Stator speed deviation of DG2 as a fault occurs at bus 684 with/without TCBR. three-phase controlled TCBR is proposed to prevent the transient instability and damp the low frequency oscillations. distribution system.02 no TCBR frequency (pu) 0 with TCBR -0.04 0.04 -0. A low cost. . THREE-PHASE DISTRIBUTION SYSTEM STUDY 129 0. The three-phase dynamic simulation results show that the TCBR can eﬀectively damp the oscillation of the whole distribution system subject to a disturbance.02 -0.06 -0.CHAPTER 11.08 0 1 2 3 time (s) 4 5 6 7 8 Figure 11.

wind turbine induction generator. selection of simulation tools. fuel cell and wind turbine DGs. 2) application of power electronics devices. Several important issues have been including the availability of simulation tools. and 3) control strategies for DGs and using FACTS when DG’s are connected to a distribution line to improve the dynamic stability. gas turbine synchronous generator and diesel engine synchronous generator have been discussed and implemented in this dissertation. Three sections of the literature are surveyed: 1) mathematical models and practical parameters for gas turbine. and to design controllers to improve the dynamic stability of the integrated power distribution system.130 Chapter 12 Conclusion The objective of this dissertation is to develop improved dynamic models for distributed generations (DG). A two-year DG project at West Virginia University (WVU) evaluated the impacts of various DG sources on the stability of actual distribution systems by computer simulation using data supplied by two regional electric utilities. modeling of various DGs and impact of DGs on the distribution system under a variety of operating conditions. The generator model of wind turbine induction generator is represented by a full order induction generator. A simpliﬁed gas turbine model is chosen to be implemented. The rating power . to investigate their impacts on dynamic stability of power distribution systems. Four types of speciﬁc DGs: fuel cell power plant. A practical diesel engine for emergency use is modeled. By focusing on these objectives the interaction among DGs has been evaluated and the dynamic stability of the integrated power distribution system has been examined. such as Static Var Compensation (SVC) and Thyristor Controlled Braking Resistor (TCBR) in stabilizing power distribution systems. The generator model of gas turbine generator and diesel engine generator is represented by a full order synchronous generator model. diesel engine.

the fuel cell plant control is the best choice for overall coordinated operation. One DG control is fuel cell control. Simulation results show that good conﬁgurations of DGs within the system can improve system stability. Dynamic load characteristic have been considered by modeling them as structured uncertainty. Among the DGs investigated in this research. Two types of fuel cell models are developed. the three-phase thyristor controlled braking resistor (TCBR) is connected to absorb surplus energy when the system is subjected to a disturbance. Finally. the other is wind turbine control. In wind turbine DG control design procedure. Linear analysis methodologies were utilized in designing controls. Coordinated operation of all the DGs is investigated. This dissertation has not included other types of DG. In fuel cell control design. the other is a dynamic model of a solid oxide fuel cell (SOFC). Multi-DG control is investigated in this dissertation. Since fuel cell plant is interfaced with the network via inverter. two pairs of critical modes have been identiﬁed using eigenvalue analysis and the participation factors. Two speciﬁc lead-lag compensation units were designed to damp each modes respectively. An output feedback controller is designed. Simulation results demonstrate the eﬀectiveness of the proposed strategy. the inverter model is developed together with the interface. The impact of the DGs is evaluated on the three-phase unbalanced distribution system. three rotor speed deviations are used as input signals while the controller outputs are the ﬁring angle for SVC and the pitch angle for the wind turbine DG. Both controls are centralized schemes. The control of wind turbine generator is accomplished through a parallel connected SVC by adjusting the reactive power injection into the system. The gains of the two compensation units were then obtained via optimal control methodology. such as solar power plants. Control of fuel cell (SOFC) plant is accomplished through inverters by adjusting its active power injection into the network during the transient period using fast acting voltage-source inverters. CONCLUSION 131 operating region is considered for impacts evaluations and controller design. the fast acting SVC is very eﬀective in improving system damping. µ-analysis is used to evaluate the robust stability of the controllers with respect to parameter uncertainties. a full three-phase approach to model a power distribution system is investigated.CHAPTER 12. The simulation results demonstrate the eﬀectiveness of the control strategies. The IEEE-13 node radial feeder with an existing gas turbine DG and an existing diesel engine DG is used as a system to test the multi-DG control. Solar . Furthermore. One is an actual fuel cell model of the phosphoric acid fuel cell (PAFC) obtained through the data ﬁtting. In the simulation. The threephase 13-node system with gas turbine and diesel engine DGs are simulated in the Power System Blockset (PSB).

interactions with other types of DGs. In a more advanced approach. the advanced three-phase representation of distribution systems needs further development. the diesel engine. a three-phase representation is implemented in PSB. CONCLUSION 132 energy has been commercially used in the electric utility industry. At present. the gas turbine. In this dissertation. no tools are available to linearize the nonlinear model in PSB.CHAPTER 12. Then. with the transformation of dq0 to abc. . and the dynamic control of solar DG could be investigated. so the approach can be used in large-scale systems. The advantage of this methodology is that the speed could be tremendously improved. these DGs models could be connected to the three-phase network. the running speed will become very slow. The major problem is to solve the three-phase unbalanced power ﬂow and to set the initial condition of each state variable. Impacts on the distribution system with existing DGs. the fuel cell and the wind turbine DGs models could be used with the zero-axis circuits added. so this three-phase approach is not suitable to design a controller system directly yet. If the number of state variables becomes very larger. Also. Dynamic modeling of solar could be considered in the further research.

Vol. F.M. 1. Gener.” IEEE Trans. Bessette II.” Chem. Battista. R. Salameh: “Dynamic Response of a Stand-Alone Wind Energy Conversion System with Battery Energy Storage to a Wind Gust.M.M. and Z. 768-773. No. Xu: “Control Design and Dynamic Performance Analysis of a Wind Turbine Induction Generator Unit. Vol.A.A. [7] H. Transm. 73-78. Power Systems. No. Revised Printin.B. pp. J. New York. Vol 145. Fouad: Power System Control and Stability. Eng. Belhomme et al: “Case Study on the Integration of Non-Utility Induction Generator to the Hydro Quebec Distribution. Wepfer: “Electricochemical and Thermal Simulation of a Solid Oxide Fuel Cell. Vol. Champagne: “Application of a Multivariable Feedback Linearization Scheme for Rotor Angle Stability and Voltage Regulation of Power Systems.A.” IEEE Transaction on Power Systems. [9] N. Vol. Borowy. pp. [6] P. pp. 14. Anderson.S. July 1995. and C. and R. pp. pp.S. Energy Conversion. Inc.” IEE Proc. The Institute of Electrical and Electronics Engineering. Mantz.” IEEE Trans. [3] O. [10] B. 2. 4.F. Puleston. Alghuwainem: “Speed Control of a PV Powered DC Motor Driving a Self-Excited 3-Phase Induction Generator for Maximum Utilization Eﬃciency. 1996. Barbosa. Okou. 15. Badran. and M. Power Systems. 5..M. March 2000. 1-15. 4. [8] R. Vol.133 References [1] E. and W. No. Watanabe.D. No. No. Vol. [4] S. Hanitsch: “Control Strategy for gridconnected DC-AC Converter with Load Power Factor Correction. 3. 147. L. Choudhry: “Design of Modulation Controller for AC/DC Power System.” IEEE Trans. Vol. Vol. 15. No.H..” IEEE Trans. No. 620-628. 1994. March 1997. May 2000. pp. 1. May 1999. Energy Conversion. and R. L. 1490-1496.F. 12. 11. Dessaint.G.G. Comm. 728-734. December 1996. Rolim.” IEEE Trans. [5] S.J.A. Christiansen: “Sliding Model Control of Wind Energy Systems with DOIG — Power Eﬃciency and Torsional Dynamics Optimization. Distrib. and W. 487-491. 2. and A. Abdin. Energy Conversion. November 1993. September 1998. E. No. Akhrif. P. .” IEEE Trans. [2] P. F. Power Delivery. 10. 8. pp.

Proceedings of the 22th Annual North American. Vol.” IEEE Trans. 14. Vol. Energy Conversion. P. Chamali: “Adaptive Fuzzy Control for Wind-Diesel Weak Power Systems.” Columbus. and J. [16] T.K. No. OH. No. and K. 11. Bussel.M. 1990. 1. and R. B. Cho: “Analysis and Controller Design of Static Var Compensator Using Three-Level GTO Inverter. 16. .C. March 2000. Vol. [19] Z. [21] J. Choudry. pp. D. [22] M. 3. No. and C. Kim: “Advanced Voltage Regulation Method at the Power Distribution Systems Interconnected with Dispersed Storage and Generation Systems. S. Brogan: Modern Control Theory. Jung. 2. Vol.B. pp. April 2000. R.” IEEE Trans. N. F. 1. Ellithy: “Design of A Robust Modulation Controller Over A Wide Range of Load Characteristics for AC/DC Systems. pp. and U. December 1995. 232-238. No. Koene. Reza. Bonanno. Energy Conversion.M. and R. 2-7. Vol.L. Chandorkar. and J. 57-65.H. Adapa: “Control of Parallel Connected Inverters in Standalone ac Supply Systems. June 2001.S. 625-635.H. Mallant: “Dynamic Model of Solid Polymer Fuel Cell Water Management. pp.F.A. June 1999. Power Delivery. 1990.A. No. [23] M. Consoli. F. [14] A.” Journal of Power Sources. June 2001. pp. Nocera: “Transient Analysis of Integrated Diesel-Wind-Photovoltaic Generation Systems. New Jersey: Prentice Hall.H. 2. Vol. and J. Zioga. and K. Vol. Pena. No.L. . M. 5. and E. Energy Conversion. pp.” IEEE Trans. [12] W. Choudry.” IEEE Trans. 2001. Raciti. No. Cardenas. 691-696.H. A.REFERENCES 134 [11] F. [13] H. IEEE PES Winter meeting. 71-78. Vol. G. Choi. 1998. pp. Chan: “Performance Analysis of a Three-phase Induction Generator Self-excited with a Single Capacitance. 15. 14. 269-274. A. Third Edition.G. 1991. Divan.G. January 1996. pp. Vol 71. May 1994. Spooner: “Grid Power Quality with Variable Speed Wind Turbines. 136-143.C.P. No.A. Lateef: “Robust Design of Dynamic Compensator for HV AC Power System”. pp. Industrial Electronics. G.C. Campos.” IEEE Trans. 218-222.” IEEE Transaction on Power Systems. Vol. 894-900. 3. No.H.” IEEE Trans. 15. Power Electronics.” IEEE Trans. February 1990. Chen.E. 9.” IEEE Trans. No.A. Lindsay: “Analysis and Design of a Series Voltage Unbalance Compensator Based on a Three-Phase VSI Operating With Unbalanced Switching Functions. Energy Conversion. [24] Distribution System Analysis Subcommittee Report: “Radial Distribution Test Feeders. Morgana. Joos. pp. Chedid. 1. Choi. January/February 1993.” IEEE Trans. 148-154.D. [18] R. 48. Industry Application. [15] R. [20] G. Vol. G. 29. Power Symposium. Karaki. Cho. 4. 1. [17] M. Clare: “Control Strategies for Enhanced Power Smoothing in Wind Energy Systems Using a Flywheel Driven by a Vector-Controlled Induction Machine. Power Electronics.A. Asher.H. 2. pp.

15. 2. Choudry: “Eﬀect of Load Models on AC/DC System Stability and Modulation Control Design. E. Vol.R. Aug.” IEEE Trans.” IEEE Transaction on Power Systems. Power Systems. 12. No. 8. May 1998.M. pp. [29] K. February 1993.W. January 2000.A. Vol. [27] M. J. Power Delivery.H.” IEEE Trans. May 1996. and M. 4. Vol.” IEEE Transaction on Circuits and Systems. 1.REFERENCES [25] Distribution System Protection Manual. 1973. No. Joshi: “A New Approach to Load Balancing and Power Factor Correction in Power Distribution System. 15. Hauer. Power Systems. Power Systems. Vol. and A. Smith. M. Gasca: “Coordinated Control of Two FACTS Devices for Damping Interarea Oscillations.” IEEE Trans. 1.G.G. [38] Electric Power Distribution System Engineering. pp. D. Vol. 417-422. PWRD-1. [35] A. Keri. January 1994. New York. Khammash. and A. [31] Electric Power Research Institute: Transmission Line Reference Book. 2. No. Vol. 741-746. 2. October 1994.B. Second Edition. Vol. Power Delivery. . A. McGraw-Hill Book Company. Feliachi: “Identiﬁcation of Critical Modes in Power Systems. and R. and M. PWRD-1. Gole.S. and G. 5.M. No.J. Vol.” IEEE Trans. 1986. K. Choudry: “Eigenvalue Assignment Using Optimal Control. and etc: “Guidelines for Modeling Power electronics in Electric Power Engineering Application. Emarah. pp. No. Adapa: “Control of A Dynamic Brake to Reduce Turbine-Generator Shaft Transient Torques. Galanos.B.” IEEE Trans. Vittal: “Sensitivity Based Structured Singular Value Approach to Stability Robustness of Power Systems. Johnson. Vol.” IEEE Transaction on Power Systems. 1252-1257. R. Donnelly. Power Systems. No. 1.F.A. May 2000. 1. Trudnowski.I. [30] A. No. 8. pp.W. 2. Brush. [34] J. 11.S. 505-514. Power Systems. C. 15. 13. and V.J. Cheng.D. [37] A. Ghosh. Dagle. X.” IEEE Trans.A. 825-830. Donnelly. [28] M. [33] G. Vol. Rogers: “Impacts of the Distributed Utility on Transmission System Stability. pp. February 1993. Vol. J. New York: McGraw-Hill Book Company. 2. April 1986. Power Delivery. No. 1982. E. Cerrada: “Control System for PWM-Based STATCOM. J.” IEEE Trans. Djukanovic.K. Gunther. 135 [26] M. [32] A. [39] P. Ellithy. 1990. Gonzalez. Hatzladoniu. pp.” IEEE Transaction on Power Systems. No. April 1986. 11. [36] W. No. Gish: “Small Induction Generator and Synchronous Generator Constants for DSG Isolation Studies. pp. Maratukulam: “Advanced Static Compensator for Flexible AC Transmission. 3. J. November 1985. No. CAS-32. No. R. 67-73. J. and D. Vol. 23.” IEEE Trans.

3.” IEE Proceedings-C. pp. 12. 22-28. July 1991. 133-140. G. Vol.P. and V.J. 342-350. No. Q. ” IEEE Trans. September 1999.REFERENCES 136 [40] Q. March 2001. Durrant-Whyte: “Fuzzy Sliding-Mode Controller with Applications. G. No. Gao. 3. No. pp. F. .F. 1035-1041. Dumas: “Dispersed generation increase the complexity of controlling protection. Vol. Vol. 39. pp. and F. . S. and B.Fardanesh: “A Governor/Turbine Model for a Twin-Shaft Combustion Turbine. 1999. No. No. [53] T. Mgmt. No. Hannett. 10. No.” IEEE Transaction on Industrial Electronics. [51] J. 2.” IEEE Trans. February 1997.L. Power System Blockset for Use with Simulink[Online]. 1. pp.H. 12. 6. 1605-1610. 48. Hingorani. Vol. Hung: “Dynamic simulation of gas-turbine generating unit. [42] D. 4.” IEEE Transaction on Power Systems. No. Computer Applications in Power.C. pp. Pai: “Design and Operation of Grid-Connected Photovoltaic System with Power-Factor Control and Active Islanding Detection.mathworks.G. Huang. [41] N. Nguyen. 9. Huang. 1998.com. pp. 1. April 1999. Vol. [47] N. June 1995. and W. Hatziadoniu. Energy Conversion. New York: IEEE Press. Understanding FACTS. [54] Hydro Quebec TEQSIM International. Vol.F.J.” Energy Conver. Energy Conversion. pp. Vol. [43] T.” IEEE Trans. pp.C. and J. and F. February 2001. pp.N. Rye. 749-753. He: “Dynamic Model for Molten Carbonate Fuel Cell Power-Generation Systems. Jee.” IEEE Spectrum. February 1993. Milanovic: “Locating Dynamic Loads Which Signiﬁcantly Inﬂuence Damping.J. Hung. T. Hammons. No. and R. pp. pp. 2-22. Ha. Lai: “Voltage Dips due to Direct Connection of Induction Generators in Low Head Hydro Electric Schemes. Vol 148. [48] N. Available: http://www. and maintaining the distribution systems. Hall.” IEEE Trans. and J. August 1991.” IEEE Trans. Canard. Gyugyi. pp. W. Colclaser: “Transient Modeling and Simulation of a Tubular Solid Oxide Fuel Cell. [52] S. Chen. [49] I.” IEEE Trans.. 2. February 1995. 255-261.V. Vol. Chalkiadakis. W. Feiste: “A Power Conditioner for a GridConnected Photovoltaic Generator Based on the 3-Level Inverter. and S. Energy Conversion. Hingorani: “Introducing Custom Power. Vol. September 1994. and H.T. 1. J.Y. Hung: “Variable Structure Control: A Survey. 243-250. Transm.” IEE Proc. Yang: “Power System Output Feedback Stablizer Design via Optimal Subeigenstructure Assignment. pp.Y. [50] W. 8. [46] W. 14. 3. Hwang. [44] L. and L. [45] C. No.” IEEE Transaction on Power Systems. 4.S. No. 38-46. 1.C. Hiskens. 362. pp. Distrib.K. 40. 460-465. Power System. No. Vol. 41-48. Vol.G.Gener. 138. 14. December 1999. 775-783.A.E. D.J. Hadjsaid. Industrial Electronics.

Energy Conversion. Engineering and Economics. [64] Y.” IEEE Trans. Vol. No. and G. Kana. pp. July 1993. Habetler. 2204-2210. “A Cost Eﬀective Generator Brake For Improved Generator Transient Response. vol. 1971. 140.S.” IEE Proceedings-C. 2000.. Vol. No. V. Power Systems Restrcuturing.H. [57] IEEE Working Group on Prime Mover and Energy Supply Models for System Dynamic Performance Studies: “Hydraulic Turbine and Turbine Control Models for System Dynamic Studies. 14. Technical Report. . West Virginia University.REFERENCES 137 [55] IEEE Distribution Planning Working Group Report: “Radial Distribution Test Feeders. Power System. 9. 3. J. No. 7. and Z. 10. pp. 4. 975-985. [67] M. [60] H. August 1991. Power Apparatus and Systems. Choudhry. 3. and D.L. 494-509. Miao. vol. 20. Kim: “An Electrical Modeling and Fuzzy Logic Control of a Fuel Cell Generation System. Bhattacharya: “Design of variable structure power system stabilizers with desired eigenvalues in the sliding mode. Vol. and L. Dorsey. 9. Power Electronics. 263-268.A. Morgantown. Kariniotakis. No.L.” IEEE Trans. and K. 2. Power Systems. Morgantown. [61] C. Final Report. Feb. Power Systems. and A. 2. 3. New York: John Wiley & Sons. Klein. Kimbark: Direct Current Transmission.” IEEE Transaction on Power Systems.” IEEE Trans. Klein. Nov. vol. Vol. F. Z. pp. pp. Choudhry. 239-244. M. 1. Jiang.W.A. No. pp. M.” IEEE Trans. [56] IEEE Special Stability Controls Working Group: “Static Var Compenstor Models for Power Flow and Dynamic Performance Simulation. 5. M. PAS-100. and S.. West Virginia University. [66] R.L. Kim. Miao: Impacts of Distribution Generation on Voltage and System Stability. May 2001. vol. April. 1. February 1981. pp. and K. “Transient Behavior of a Class of Wind Turbine Generators During Electrical Disturbance. 1994. pp. Kluwer Academic Publishers.” IEEE Trans. [59] M. Vol. 375-381.C. Fan: Impacts of Distribution Generation on Voltage and System Stability. No. and L. Energy Conversion. Nanda. Fink. [68] P. PART II: Implementation of the Algorithm and Case-Studies With Induction Generators.” IEEE Trans.N. [65] R. June 1999. Galiana. [62] G. Krause. Stavrakakis: “A General Simulation Algorithm for the Accurate Assessment of Isolated Diesel-Wind Turbines Systems Interaction.L Kothari. [58] IEEE Working Group on Computer Modelling of Excitation Systems: “Excitation System Models for Power System Stability Studies. No. T. February 1992.S. Ilic.” IEEE Trans. No. 6. 2001. 16. Vol. Vol. pp. 1994. Wolf: “System Management of a Wind-Energy Converter.T.” IEEE Trans. Eckroth. 584-590. [63] E. 1998. Thamodharan. PAS. Boston. 1. Man. September 1995. May. J. PAS-100. 1840-1846. pp. May 1981. No.

[77] M.J. and V. Vol. November 1996. [80] Z. L. 4. and S. [71] X. Lin. Power Systems. .D. Y. and R. Vol. February 2000. No. [75] W. No. College Station. Mello. 299-306. Lalk. O. Chin. TX. Klein. T.C. and R.N. [82] Z.A. L. August 2001.REFERENCES 138 [69] P. 473-480.” IEEE Trans. [72] F. 4. Power Systems. Choudhry. PAS-101. 16. [81] Z. and D.G. No. Choudhry.C. No. Syrmos. Vol. Klein. Choudhry: “Case Study on the Impacts of Distributed Generation on the Power Distribution System.” IEEE Trans. and L.R.G.” IEEE Trans. Vol. R. Lukas. Lerch. December 1999.” Journal of Power Sources 70. 1. 11. Optimal Control. pp. 16. 1998. Vol. and D. Lukas. August 1999. October.D. September 2001. pp. Lewis. Ayagh: “Development of a Stack Simulation Model for Control Study on Direct Reforming Molten Carbonate Fuel Cell Power Plant. 9. 15. New York City. 3. Miao. E.H.L. Inc. 14. No. PAS. No.D. 289-295.L. K. Second edition. 4. 3385-3393. Teng: “Three-Phase Unbalanced Distribution Power Flow Solutions with Minimum Data Preparation. H. No. pp. and M. Krause. and J. Hannett. and A. 5. Miao. Vol. 1178-1183. 1995.” Proceeding of North American Power Symposium 2001. pp. New York. White. and H.. 258-268. November 1990. Analysis of Electric Machinery. M. Lee. vol. Vol. 31.C. Miao. [79] F. Appleby: “Modeling Electrochemical Performance in Large Scale Proton Exchange Membrane Fuel Cell Stacke.” submitted to IEEE Trans. K. Power Systems. IEEE Press.” IEEE Trans.L. “Dynamic Model and Control of A Fuel Cell Power Plant in Power Distribution System. Povh: “Optimization and Coordination of Damping Controls for Improving System Dynamic Performance.L. Power Systems. Martins. Energy Conversion.” IEEE Trans. 14. Fernandes: “Assessing Eigenvalue Sensitivities. Klein. Energy Conversion. Power Systems.E.” IEEE Trans. Lima: “Determination of Suitable Locations for Power System Stabilizers and Static VAR Compensator for Damping Electromechanical Oscillations in Large Scale Power Systems. Feltes.M. [78] N. “Application of Induction Generators in Power System. and J. 1455-1469.J. 2001.T. No. Karlsson: “Distribution System Modelling for Voltage Stability Studies.W. 26-Jan. [70] J. Lima. 1995.N. September 1982. [73] E. Lind. 3.Y. Wasynczuk. pp.S. Lee.F.” IEEE Trans. J.” IEEE PES Winter Meeting 2002.A.D.S. 1651-1657.” IEEE Trans. and H. Power Systems. [74] R. 3. M. pp. [76] M.H.Y. pp. Su. Lee. New York. Jan. pp. “Dynamic Simulation and Stability Control of Three-Phase Power Distribution System With Distribution Generators.A.L. Ayagh: “An Explicit Dynamic Model for Direct Reforming Carbonate Fuel Cell Stack. Sudhoﬀ. John Wiley & Sons. Lei.

Tempe. 10. Mohan. 4. Asher: “Doubly Fed Induction Generator Using Back-toBack PWM Converters and Its Application to Variable-Speed Wind-Energy Generation. Murdoch. No. 220-229. 3. R. pp. Miao.A. Power Systems. Power Apparatus and Systems. February 2000. Barton J. Vol. 10. 1920-1926. Kim. G. Milanovic. May-June 2000. Wind-electric. Hiskens: “Damping Enhancement by Robust Tuning of SVC Controllers in the Presence of Load Parameters Uncertainty. 2000. [91] T. pp. May 1999. and T. Vol. Nam. Power Delivery. AZ. Power Delivery. 15. J. 231-241.” American Scientist. 1. 15. No. Bulawka: “A Novel Grid Interface. [94] B. Optimized for Utility-Scale Applications of Photovoltaic. 1298-1303. M. T. No.R.Y. Power Systems.A. October 1995.” IEEE Transaction on Power Systems.” IEEE Trans. pp. May 1983. Miao.0. Vol. pp. M. [96] R. No.R. No. [86] T. 2. Applications. Tempe. No.” IEEE Trans. 2020-2026.S.W. and I. 1998. [89] R.” accepted by North American Power Symposium 2002.. AZ. McDond: “An integrated SOFC plant dynamic model for power systems simulation. Kihara.R. Lee: “A New Eigen-Sensitivity Theory of Augmented Matrix and its Applications to Power System Stablity Analysis. and Design.A. Winkelman. PAS-102. N. Hiyama. Mohan. pp. [88] A. [84] Z. [85] J. Naik.K. vol. No. 1. 129-136. Inc.H.F.K. Rogers. Pena.K.H. and J.J. [92] J.S. 363-369. Shim. Electr. M. [87] N. 88. November 1998. Power Appl. R. 13. Vol. and S. Robbins. Vol.” accepted by North American Power Symposium 2002.” IEEE Trans. Iravani: “Dynamic Modeling of High Power Static Switching Circuits in the dq-frame. 1995.R.A. and L. [95] Power System Toolbox-Dynamic Tutorial and Functions. and Fuel-cell Systems. Acha. H. [90] H. May 1996. pp. Perkins. K. 14. Version 2. [93] H. Perez.L. “Control of Wind Turbine-Generator in a Power Distribution System with Multiple Types of Distributed Generators (DG).” IEEE Trans.L. . pp. Power Electronics: Converters.” IEEE Trans. No.P.Mishiro. October 1995. February 2000. Choudhry. and C. “A Fuel Cell Model from Data Fitting Method and Its Impacts on a Power Distribution System with DGs. Undeland. 143. 4. Fan.V. and R. Klein. Choudhry.M. and A. 5. 4.” IEEE Trans. Ault. Vol.M. Ontario Canada. Clare.C. Vol.” Journal of Power Sources. Power Systems. 1340-1347. pp. New York: John Wiley & Sons. 495-500. Padulles.REFERENCES 139 [83] Z. an K. Ortmeyer: “Fuzzy Logic Switching of Thyristor Controller Braking Resistor Considering Coordination with SVC. M. Y. E. Overbye: “Reengineering the Electric Grid. Klein. M. Cherry Tree Scientiﬁc Software. and W. Vol. and G. pp. Esquivel: “Advanced SVC Models for Newton-Raphson Load Flow and Newton Optimal Power Flow Studies.” IEE Proc. Javid: “Control Design and Performance Analysis of a 6 MW Wind Turbine-Generator.

” IEEE Trans. and S. Power Delivery. Vol. 11. 105. and M.S. 2. Devices. Rahim. Rau. pp. 36. Vol. September 1998. Long.A.F. 1077-1082. Yang: “STATCOM Control for Power System Voltage Control Application. 3.” IEEE Trans. No.M. 5. No. 3. 2014-2020. pp. 3. 4. pp. May 1984. 532-541. Vol.: “Application of STATCOMs to Wind Farms. Power Systems. A. pp. Power Apparatus and Systems. August 1988. [103] N.” IEEE Trans. No. [109] S. No. 4. [110] Z. 986-991. 1311-1317. Vol. Wan: “Optimum Location of Resources in Distributed Planning. pp.” IEEE Trans. pp. Thorp: “An Adaptive Optimal Control Strategy for Dynamic Stability Enhancement of AC/DC Power Systems.H. A. and V. Power Systems. pp.” IEEE Transaction on Power Systems. May 1993.M. February 1996. Vol. Vol. [102] Power Electronics: Circuits. October 2000. November 1994. W. No. 3. No. and Y. Ganesan: “Coordinated Local and Centralized Control in Distribution Management Systems. 865-869. [100] C. 1998. E. [107] W. No. and Z. Industry Application.” IEEE Trans. and D. Rubira. October 1983. and A. 4. Vol.D. Rao. Rahim. E. J. [98] A. Power Electronics.” IEEE Trans. Sarkar: “Half Wave Thyristor Controlled Dynamic Brake to Improve Transient Stability.J.S. 145. July/August 2000. 15.H. 13. Vol.L.” Journal of Engineering for Power.K. PAS-103. 511-516. Vol. Second Edition. . M. M. 15. Vol. Vol. 879-886.L. and J. 718-724. 422-427. 2. 11.B.” IEEE Trans.” IEEE Trans. [108] I. Rao. M.M. Rahman: “Two-Loop Controller for Maximizing Performance of a GridConnected Photovoltaic-Fuel Cell Hybrid Power Plant.H. Al-Shehri. [104] D. Al-Shehri: “Optimum Control Stategies for Transient as Well as Oscillator Instability of Power Systems. 9.” Vol. and etc. April 2000. No. 8. 491-496. [106] K. Rostomakolai. pp.N.” IEEE Trans. 5.A.A. Roytelman. . Rahim. T. and Application. No. McCulloch: “Control Method Comparison of Doubly Fed Wind Generators Connected to the Grid by Asymmetric Transmission Lines. Ro. September 1998. Power Delivery. Reichelt. pp. Saad-Saoud. No.G.I. Power Systems. Vol. New Jersey: Prentice Hall. pp. Alamgir: “A Closed-loop Quasi-Optimal Dynamic Braking Resistor and Shunt Reactor Control Strategy for Transient Stability.A. Ekanayake. Lisboa. pp.D.REFERENCES 140 [97] A. No. Energy Conversion. 276-281. Crow. Phadke. Power Systems. and H. 1. Quaicoe: “Analysis and Design of a Multiple Feedback Loop Control Strategy for Single-Phase Voltage-Source UPS Inverters.I. 4. Rowen: “Simpliﬁed Mathematical Representations of Heavy-Duty Gas Turbines. No. August 1988. Echnauer.S.H. J. Glavitsch: “Estimation of Steady-state Unbalanced System Conditions Combining Conventional Power Flow and Fault Analysis Software. 3. pp. [99] N. [101] P.” IEEE Trans. July 1996. [105] N.

5. [114] C. 1999.” IEEE Trans. 1.” IEE Proc. I. No. 26-29. [123] M.” 34th Universities Power Engineering Conference. [115] J. IEEE International Conference on. UK. pp. May 1999. Vol. and V. No. No. 4. 2122-2125. 11. IA-22.” IEEE Trans. 820-827. Noguchi: “A New Quick-Response and High-Eﬃciency Control Strategy of an Induction Motor. No. 1978. Power Electronics. [122] I. D. 16. B. Vol. 1. 14. Tsang: “Improved probabilistic method for power system dynamic stability studies.J. 259-264. Power Delivery. March 2001. pp. She. 3. Asiain. 971-978. 147. pp. 2. Energy Converison. Vol. K. Jr. K. C. Tsuji.R. No. No. pp. pp.” System Engineering.” IEEE Trans. Ramanarayana: “Analysis and Performance Evaluation of a Distribution STATCOM for Compensating Voltage Fluctuations.REFERENCES 141 [111] O.” IEEE Trans.W. 291-294. Leicester. No.S. Elements of Power System Analysis. [118] J. Samuelsson. 1982.-Gener. [124] K.D. Mitani. Olguin: “The eﬀect of induction generators in the transient stability of an electric power system. 172-181. vol. and O. Santos. pp. 9. Vol. Vol. 608-613. 2. Shinoki. and T. Sensarma. Sobczyk. pp. Distib. Second Edition. April 2001. 1991. and M.” IEEE Trans. 722-729. Stevenson. Y. Gjosater: “Robust Control and Analysis of a Wind-Diesel Hybrid Power Plant. AddisonWesley Publishing Company. Applied Superconductivity. Power Delivery. January 2000. December 1994. Energy Conversion. M. Transm. Univ. Fourth Edition. [116] T.M. Akke: “On-oﬀ Control of an Active Load for Power System DampingTheory and Field Test. 701-708. pp.M. January 2001. [119] T. 2. and D. Wang. No. Modern Control System Theory and Application. Tsuda. Vol. pp. August 1998. Slotine. 37-43. and B. September/October 1986. Sapinski: “Analysis of Phase-Controlled Converters for Induction Motors. 1991.. Takahashi. Vol. 10. [120] W. W. New Jersy: Prentice Hall. pp. Ruiz.A.Y.B.J. T. 5.” IEEE Spectrum. No. . Chung. Feliachi: “On Eigenvalue Ranking Indices for Continues Linear Time Invariant Systems. Sasaki: “Development of an Internal Reforming Molten Carbonate Fuel Cell Stack. Matsumura.” IEEE Trans. pp. Uhlen. April 1990. [117] S. 4. Sharma: “Modeling of an Island Grid. 1. Vol.L. 84-86.” IEEE Trans. [112] E. and A. New York: McGraw-Hill Publishing Company. 13. Vol. [113] P. and K. and A. Shinners. [121] W. Applied Nonlinear Control. Sweet: “Networking Assets. Kakihana: “Application of Resistor Based Superconducting Fault Current Limiter to Enhancement of Power System Transient Stability. and K. December 1995. Industry Application. Li.” IEEE Trans. pp. Power Systems. Padiyar. [125] K. 1991. Foss.

pp. Power Systems. Yang. 1557-1562. 1998. Yilmaz. 4.REFERENCES 142 [126] K. Farmer: “Dynamic Voltage Compensation on Distribution Feeders using Flywheel Energy Storage. W. No.R. 2. and Y. 4. Vol.G.R. No. Vol. Yoshihara. April 1999. and H. December 1999. [138] R. Oct. H.” IEEE Spectrum. 2045-2052.” Journal of Materials Science Letters. Chung. [129] R. and F. Vol.D. and R. [128] Yu Wang. pp. New Jersy: Prentice Hall. R.C Doyle. pp. and I. Wolk: “Fuel Cell for Homes and Hospitals. Chiang: “Fast Decoupled Power Flow for Unbalanced Radial Distribution Systems. Vol. Power Systems. 3. Yang. 45-52. S. [137] K. Maratukulam: “Variable-Structure Braking-Resistor Control in A Multimachine Power System. Wang. Cho. C. L. Karady. No. [133] K. 1. Y. 465-471. H. Vol. Energy Conversion. J.” IEEE Trans. 128-137. 494-502.: “Stabilization of Inter-Area Oscillation Modes Through Excitation Systems. “Impedance Characteristics of a Phosphoric Acid Fuel Cell. and J. G. and etc. Zhou.” Simulation. 887-893.” UPEC ’97. Yeager.S. pp. pp.” IEEE Trans. pp. Tsang: “Multimachine Eigenvalue Sensitivities of Power System Parameters. 15. 254-260.Y. No.: “Integration of a StatCom and Battery Energy Storage. . and J.G. 1755-8. March 2000.” IEEE Trans.D. pp.W. 10. pp.H. Power Systems. May 2000. No. Mittelstadt.” IEEE Trans. Deng: “Transient Performance of an Isolated Induction Generator under Unbalanced Excitation Capacitors. No. Vol. Jang. May 2001. and R. Sabuncuoglu: “Input Data Analysis Using Neural Networks. [130] R. Essentials of Robust Control. Zimmerman. Power Systems. [131] X. [134] A. Mohler.” IEEE Trans. 9. [127] L. Vol. No. pp. [136] M. Vol. 2. 3. Willis: “Modeling of Emergency Diesel Generators in an 800 MegaWatt Nuclear Power Plant.J. Uriu. and K. and A. Wang.M. Power Systems. 8. August 1994. No.Y. pp. Vol. Power Delivery. 433-441.” IEEE Trans. 741-747. 16. May 1999. Weissbach. 17. and D. February 1994. 1998.” IEEE Trans. No. 14. 20. [132] Z.” IEEE Trans.A. Feliachi. November 1995. September 1993.E. Koyanagi: “Optimal Allocation and Evaluation of the Dispersed Fuel Cell Generator on the Power System. Energy Conversion. Shen. 14. [135] K. C. 2. pp. Zhang. 9. Yoon.

rotor and stator quantity.4) (A. a full-order model [69] is simulated. leakage and magnetizing inductance.5) .7). The following equations are represented in the rotor reference frame. k: d and q axis quantity. The equations (A.7) represent the 7th order state space electrical part of the synchronous generator. s: l.1 Full-Order Synchronous Generator Model To investigate the dynamic stability characteristics of a synchronous generator.1 .3) (A.143 Appendix A Conﬁguration of Models A.1-A.3 show the equivalent circuits. q: r. ﬁeld and damper winding quantity. The voltage equations are: r vqs = −rs ir + ω r λr + pλr qs qs ds r vds = −rs ir − ω r λr + pλr qs ds ds 0r 0 vf d = rf d i0rd + pλ0rd f f 0r 0 vkd = rkd i0r + pλ0r kd kd 0r 0 vkq1 = rkq1 i0r + pλ0r kq1 kq1 (A. The mechanical system are illustrated as equations (A.2) (A.A.9. Figures A. In this DG research. This model is used in the gas turbine DG and the diesel engine DG. m: f . d.A. The substation is represented as a simpliﬁed synchronous generator and will be discussed in the following section.1) (A.10). In the equations. the full-order synchronous generator ingored the zero-axis part (A.

The ﬂux linkage equations are illustrated as: λr qs r λds λ0rd f λ0r kd 0r λkq1 λ0r kq2 λ0s ir qs r ids i0rd f i0r kd 0r ikq1 i0r kq2 i0s (A.7) v0s = −rs i0s + pλ0s where p = d dt .1: Q-axis equivalent circuit of a synchronous generator. 0r 0 vkq2 = rkq2 i0r + pλ0r kq2 kq2 (A.6) (A.8) where = L_syn ∗ .APPENDIX A. CONFIGURATION OF MODELS r'kq2 rs + 144 + v'rkq2 - λrds ωr L'lkq2 Lls i'rkq2 L'lkq1 vrqs - irqs r'kq1 i'rkq1 + v'rkq1 - Figure A.

L_syn = −Lls −Lmq 0 −Lmd −Lmd 0 0 0 −Lls −Lmd 0 0 −Lmq −Lmq 0 0 Lmd L0 d +Lmd lf Lmd 0 0 0 0 Lmd Lmd L0 +Lmd lkd 0 0 0 Lmq 0 0 0 L0 +Lmq lkq1 Lmq 0 Lmq 0 0 0 Lmq L0 +Lmq kq2 0 0 0 0 0 0 0 Lls 0 The mechanical system is: .3: Zero-axis equivalent circuit of a synchronous generator. CONFIGURATION OF MODELS r'kd rs + vrds i'rfd irds Lmd λrqs ωr L'lkd Lls i'rkd L'lfd v'rkd - 145 + r'lfd + v'rfd - Figure A.APPENDIX A.2: D-axis equivalent circuit of a synchronous generator. rs + v0s i0s Lls Figure A.

9) (A. CONFIGURATION OF MODELS 146 Iq xq-x' q 1 1+sT'q0 E' d Id xd-x'd + ∑ 1 + 1+sT'd0 E' q EFD Figure A. In this model.4: Diagram of a simpliﬁed synchronous generator model. Figure A. pω r = − 1 (Te − Tm ) 2H (A. The state equations of the electrical part are: 0 pEd = 0 pEq = 1 0 (−Ed + (xq − x0 )Iq ) q t0 q0 (A.APPENDIX A.11) (A.12) 1 0 (−Eq + EF D + (xd − x0 )Id ) d t0 d0 The electrical torque output is: 0 0 Te = Ed Id + Eq Iq − (x0 − x0 )Id Iq q d (A. the synchronous machine has two stator circuits and two rotor circuit.13) The state equations of the mechanical part are: .4 shows the two-axis model.10) pθr = ω r The torque in per unit is: 3 P Te = ( )( )(λr ir − λr ir ) qs ds ds qs 2 2 A.2 Simpliﬁed Synchronous Generator Model A simpliﬁed [2] synchronous generator model is considered to represent the substation of the distribution system.

3 Induction Machine A full-order induction machine includes the 4th order DQ-axis state equations (see Figure A.5A. Also. 1st order electrical torque and 2nd order zero-axis state equations (see Figure A.17) .15) A. The electrical part is represented as 6th order state space model (see equation A. the induction machine is simulated as a 5th order model and the 2nd order zero-axis equations are ignored. this model is used to represent the dynamic load model when the mechanical torque is positive.6: D-Axis equivalent circuit of an induction machine.6). CONFIGURATION OF MODELS 147 pω = 1 (Tm − Dω − Te ) tj pδ = ω (A.16) (A. In this dissertation research.5: Q-Axis equivalent circuit of an induction machine.14) (A.7) [69].16-A. The electrical part equations of the induction machine are: vqs = rs iqs + ωλds + pλqs vds = rs ids − ωλqs + pλds (A.APPENDIX A.21). and the mechanical part is a 1st order model. rs + vds ids λqs ω + Lls L'lr λ'qr(ω-ωr) + i'dr r'r + v'dr - M Figure A. rs + vqs iqs λds ω + M Lls L'lr λ'dr(ω-ωr) i'qr + r'r + v'qr - Figure A.

APPENDIX A.18) (A.20) (A. 0 0 vqr = rr i0 + (ω − ωr )λ0 + pλ0 qr qr dr 0 0 vdr = rr i0 − (ω − ω r )λ0 + pλ0 qr dr dr (A. The mechanical part are represented in per unit as: pω r = ωb (Te − Tm ) 2H (A. and the ﬂux linkage is: Lls + M 0 M 0 0 0 λqs iqs (A.23) .21) vos = rs i0s + pλ0s 0 0 vor = rr i0 + pλ0 0r 0r where p = d dt .7: Zero-axis equivalent circuit of an induction machine.19) (A. CONFIGURATION OF MODELS rs + i0s v0s Lls i'0r L'lr r'r + v'0r - 148 Figure A.22) and λds λ0 qr = L_ind ∗ 0 λdr λ0s λ0 0r 0 Lls + M 0 M 0 0 M 0 ids i0 qr i0 dr i0s i0 0r 0 M 0 L0 + M lr 0 0 0 0 0 0 Lls 0 0 ω b is the base electrical angular velocity used to calculate the inductive reactances.24) L_ind = L0 + M lr 0 0 0 0 0 0 0 L0 lr (A.

The inverter can be a PWM voltage source inverter which uses the hysteretic current control. A. it is convenient to express the voltage and ﬂux linkage eqations in terms of reactances.8.APPENDIX A. the DC/AC inverter is used.8: Diagram of a PWM current source inverter in a distribution system. This system is simulated in Simulink. ~ Distribution system load Vfc Hysteresis current controller Current reference calculation pc qc Figure A. The PST model is used to calculate the single phase power ﬂow and . A.25) Since machine and power system parameters are nearly always given in ohms or percent or per unit of a base impedance. This controller can regulate the active power and reactive power by tracking the currents at certain references. power system toolbox(PST) and power system blockset (PSB). CONFIGURATION OF MODELS and the torque equation is expressed as: Te = λds iqs − λqs ids 149 (A. The general control diagram [6] is shown is Figure A.5 Test Systems IEEE-13 node test feeder is used to study the models and controller of the DGs. The model simulated is mainly used to investigate the dynamic characteristics of the DGs and design the dynamic stability controller.4 Voltage Source Inverter To connect the fuel cell to the distribution system.

APPENDIX A.2 Line Segment Data Conﬁg. CONFIGURATION OF MODELS 150 Subst at ion 650 646 645 632 633 634 611 684 671 692 675 680 652 Figure A. 601: 1.1 IEEE-13 Node Test Feeder The IEEE-13 node test feeder [24] is used as the test system to investigate the dynamic characteristics of the distribution system with DGs.5017 0.4236 Z= 0.0179 0. thus the substation transformer is 138kV/4. In the test. A.0478 0.3465 + j1.3375 + j1.9: One line diagram of IEEE-13 node radial feeder.3414 + j1.3849 0. Z (R +jX) in ohms per mile 0.0348 2.1560 + j0. the connected transmission system is simulated as 138kV system. The PSB model is used to test three-phase unbalanced and verify the designed controller.1535 + j0.5. A.5.16kV. the values are used to initialize the state variables in Simulink model. B in micro Siemens per mile .1580 + j0. Figure A.9 shows the test system with the two DGs.

0817 −1.7526 + j1.2112 2.6658 . Conﬁg. 603: 1.APPENDIX A. B micro Siemens per mile in 5.4236 0. Z (R +jX) in ohms per mile 0 0 0 Z= 1.3849 0.2595 5.3238 + j1.2066 + j0.1: Line segment data of IEEE 13 Node A Node B Length(ft.9597 Conﬁg.1814 0.1983 0. B micro Siemens per mile in 0 0 0 B= 4.3471 0.9958 −1.3294 + j1.1795 −0.3569 2.5017 Z= 0.4591 1.1535 + j0.1580 + j0.4246 Conﬁg. 603 602 XFM-1 603 601 607 601 604 601 Switch 605 606 B= 6. Z (R +jX) in ohms per mile 0.7436 + j1. CONFIGURATION OF MODELS 151 Table A. 602: −0.) 632 645 500 632 633 500 633 634 0 645 646 300 650 632 2000 684 652 800 632 671 2000 671 684 300 671 680 1000 671 692 0 684 611 300 692 675 500 node test feeder.6386 1.7097 −0.8999 4.6905 B= 5.6990 −1.6588 5.2998 −1.7475 + j1.7417 5.1560 + j0.

CONFIGURATION OF MODELS Conﬁg. Z (R +jX) in ohms per mile 0.3238 + j1.7982 + j0.4041 0.8999 B= 0 0 4. Z (R +jX) in ohms per mile 1.4591 Z= 0 0 1.APPENDIX A. Z (R +jX) in ohms per mile 0 0 0 Z= 0 0 1.3475 2.2066 + j0.8897 Conﬁg.8897 0 96.5193 Conﬁg. 606: 1.0328 0.2849 − j0.7097 Conﬁg.3471 2.3292 + j1.6658 0 −0.4463 0. 605: 1. 604: 1.7982 + j0. B micro Siemens per mile in 96. B micro Siemens per mile in 4.3192 + j0.6231 0 0 Z= 0 0 0 2.7891 + j0.0328 0.0143 Z= 0.3294 + j1. B micro Siemens per mile in 0 0 0 B= 0 0 4. Z (R +jX) in ohms per mile 1. B in micro Siemens per mile 152 . 607: 1.4925 + j0.3192 + j0.8897 0 0 B= 96.3569 0 0.4463 2.

74◦ 0 0 451.93◦ 151.7∠-17.23◦ 78.4∠103.13◦ 2402∠-5.58◦ 2580∠-121.7806 0 0 shown in Table A.32◦ 60.92◦ 2434∠117.72◦ 66.93◦ 2329∠115.05◦ 65.07◦ 528.4∠-157.73◦ 2426∠117.36◦ 272. Table A.99◦ 0 .94∠-57.71◦ 0 0 0 I-Phase C 569.6∠-36.62◦ 0 64.74◦ 270.8∠95.84◦ 555.18◦ 117.3∠-128.05◦ 2328∠116.29◦ 69.07◦ 170.6∠117.59◦ 2409∠-5. The three-phase power ﬂow is 97.92∠-35.8∠-2.00◦ 2334∠115.78◦ 0 I-Phase A 533.13◦ I-Phase B 438.79◦ 0 0 0 V-Phase C 2433∠117.18∠122.08∠57.11∠82.4∠-121.4∠-22.54◦ 409.9∠-4.23◦ 66.APPENDIX A.56◦ 69.23◦ V-Phase B 2555∠-121.77◦ 2592∠-121.02∠-39.14∠123.18◦ 201. Line 650-632 632-633 633-634 632-645 645-646 632-671 671-692 692-675 671-684 684-611 684-652 V-Phase A 2478∠-2.21◦ 63.14◦ 2340∠116.27∠-4.30◦ 2470∠-2.21◦ 0 2395∠-5.56◦ 224.7∠-138.25◦ 2433∠117.06◦ 2552∠-121.24◦ 198.13◦ 2411∠-5.63◦ 69.8∠-142.63∠-57.49∠-122.23◦ 284.05◦ 2340∠116.19∠58.19◦ 2550∠-121.77◦ 2580∠-121.7∠111.50◦ 2555∠-121.41◦ 53.23◦ 0 0 2411∠-5.2.6∠81.73◦ 689.1∠113.5∠-156.71◦ 71.8∠-20. CONFIGURATION OF MODELS 153 B= 0 0 0 The system has unbalanced spot and distributed loads.2: Three-phase current and voltage of distribution system without DG.

Sbase = 20. Ts =0. x0 = 0.0. T3 =0.6.01. 61% conductivity) 4) Distributed Generation Generator: T0 =0.1.1 shows the parameters for the test system simulation in Chapter 4. T0 =1. VRmax =1.0MVA.1 Parameters for SI system in Chapter 3 1) Per unit system Vbase = 23kV.05. xd =2.149+j0. H = 600s. Tmax =1. hard-drawn. xd =xq = x0 = T0 =T0 =0. This test system is also used in Chapters 5-10. TG =0.5.4283 ohm/mile. H = 9.263.0MVA.105.8049 W 2) Substation Generator: Sb =129MVA. x0 = 0.05. (Aluminum conductor. TA =0.0 s. DG transformer: Sb =4. x0 = 0.1. Turbine: 1/R=24.5MVA.1. TB =TC =0.2 Parameters for the Test System in Chapter 4 Table B.0.5. . Rt +jXt = 0.05 + j0. q0 Exciter: KA =38.3.154 Appendix B Parameters for Simulation B.05 (self base) Sb =4. VRmin =-1. q q0 d d0 Transformer: Sb =42. xq =1.1 (self base) 3) The distribution line R+jX = 0. q d d0 B.0MVA.0. Zbase = 25. Zt = 0.

SOFC N0 = 384 tH2 = 26.126 B.96 D = 0.8958 xq = 0.2 Tmin = 0 T2 = 0.16kV Substation xd = 0. PARAMETERS FOR SIMULATION 155 Table B.43e−4 tH2O = 78. .4 Parameters for the Wind Turbine DG Table B.8125 0 = 0.2 Pmin = −0.00178 Xlkq1 = 0.5 Parameters for TCBR Simulation in Chapter 11 Since the TCBR simulated in power system blockset (PSB).1198 q H = 2364 Xq = 1.8 0 rkd = 0.05 Tv = 0.91 KO2 = 2.52e−3 r = 0.003 Xls = 0.1 Rs = 0.2: Parameters for the SOFC simulation.4 Diesel Engine K = 4.01 T4 = 0.000929 Xd = 1.81e−4 tO2 = 2. B.05 Tf 1 = 0.0M V A Vbase = 4.6 Gas Turbine Kd = 25 Tg = 0.2 shows the parameters for the SOFC simulation in Chapters 6 & 7.1: Parameters for the test system simulation.0384 B.08125 Xlkd H = 5.009 x0 = 0.1 T3 = 0.8645 r=0 Td0 = 8.8 0 rf d = 0.1198 d Tq0 = 0.02 T6 = 0.19 0 0 rkq1 = 0.01334 Pmax = 1. the paramters of a gas turbine DG and a diesel engine DG are based on the Pb1 and Pb2 respectively.25 T5 = 0. Base values Sbase = 4.0 T1 = 0.3 shows the parameters for the wind turbine DG simulation in Chapters 8 & 9. Table B.3 Parameters for the SOFC Simulation Table B.3 Ls = 0.APPENDIX B.1 x0 = 0.01 KH2O = 2.1 Synchronous generator rs = 0.9 KH2 = 8.

035 R=2.1 XM =4.96m Vrated =9.0 TCBR R0 =12Ω X0 =1.173 T´d =1.476 KR =33.6 T´´d =0.1MW Xd =2.2 Tmin =0 Ke0 =1. PARAMETERS FOR SIMULATION 156 Table B.001 Kp =0 .035 Pmin =-0.009 Excitation System Tr =20e-3 Ka0 =200 Te =0 Tb =0 Tf =0.APPENDIX B.00417 Td =0.1Ω Vb =4160V X´´d =0.0 T1 =0. Synchronous Generator Pb1 =2.2 SVC XC =0.161 rs =0.2 Xq =1.3m/s Ki =6.0059 ρair =1.02 T6 =0.01 T4 =0.4 Diesel Engine K=4.05 Tv =0.002 Tc =0 Ef min =0 T3 =0.075 r´r =0.3: Parameters of the wind-turbine DG.225 Tb =0.03 X´d =0.5 XL =0.1 T2 =0. Induction machine Xls =0.0384 Ta =0.0 Kf 0 =0.0339 H=0.135 X´lr =0.3 TR =0.4: Parameters for the TCBR simulation.6 Pmax =1.85e-3 Gas Turbine Kd =25 Tg =0.0014 Table B.18 X´´q0 =0.1 Ef max =6.05 Tf 1 =0.2 Xl =0.25 T5 =0.5 Wind-turbine R=69.1MW Pb2 =1.26 X´´q =0.2 Kp =6.105 H=2.

The dynamic loads are represented as induction motors. Figure C. power distribution network models.2. In Figure C. The Simulink workspace is used as running environment. . so they can be considered as the DGs absorting the real power. PEj is the power electronics device. wind turbine. full-order synchronous generator model. etc. measurement units and analysis tools. During the dynamic simulation procedure.m ﬁles. It has a different program structure. The numerical calculation methodologies are also provides by Simulink.3. induction machince model. solid oxide fuel cell (SOFC) model. a software package is developed in Matlab/Simulink. These functions are written in *. and the network includes the distribution lines. and static var compensator (SVC) model. As an example. phosphoric acid fuel cell (PAFC) model. C.1 Models The models in this software package include: transient synchronous generator model. transformers and static loads. Each of the models are connected in Simulink.1 shows the general structure of the simulation system. An example of the general simulation structure in Simulink is shown in Figure C. This package includes: DG models. This system is used to study the wind turbine DG in Chapters 8 & 9. diesel engine. the SOFC model is illustrated in Figure C. the Matlab functions are called. current source inverter (CSI) model. The three-phase approach is implemented in power system blockset (PSB). the DGi is the distributed generation.1.157 Appendix C Program Structure of Simulation In this dissertation research. gas-turbine.

Id. (δ) .q To each Sub. Id.q DGi Vd. (δ) Reference Frame Transfer Vx.2 Clock Vx [a4] From4 em interf ace Vy [a4] Goto4 ind_svc_group Vx [a5] Goto5 [a5] From5 em Interf ace Vy mot_com_group Figure C.q P E1 Vd...q.1 em Vq Interf ace Vd [a3] From3 [a3] Goto3 Diesel-engine Syn Gen. DG..1: General structure of the simulation system. . (δ) .q Subst aion Vd.y Reference Frame Transfer Id. PROGRAM STRUCTURE OF SIMULATION 158 Id.q.q.y Network Ix.q DG1 Vd.2: General structure of the simulation program in Simulink.. (δ) Figure C. δ Id. IEEE 13 node feeder with DGs (3) by Zhixin Miao Nov. or PE Id. 2001 Substation + Gas turbine + Diesel engine + wind turbine +svc +dynamic load [a1] From1 em Vq interf ace Vd [a1] Goto1 Substation Vq Interf ace Vd [a2] From2 em [a2] MATLAB Function Network em Goto2 Gas-trubine Syn Gen.q.q.APPENDIX C. 10.q P Ej Vd.

V is the voltage of each node. All the models are simulated in rotor reference frame (d. it is also used to represent the dynamic load. I =YV (C. The transient synchronous generator is to represent the substation. 5th induction machine model is connected to the wind turbine to represent wind turbine DG. I and V are complex quantities. Y is the network matrix.2 Network The network is represented by algebraic equation. C. I and V are seperated as real and image quantities.s+1 Fcn1 1/KH2 O tH2O. diesel engine DGs. Output of each model are the voltage and input is the current.3: Simulation of a SOFC in Simulink. PROGRAM STRUCTURE OF SIMULATION 159 Step I0 Co nstant3 1 qH2 q1 Co nstant4 Ga i n -K 2 qO 2 Ga i n3 0. .s+1 Fcn2 1/KO2 tO 2.1) where I is the input current of each node. diesel engine to represent the gas-turbine.s+1 Scope 8 p1 Co nstant p2 Co nstant1 p3 Co nstant2 Scope 3 Scope 6 V-I f(u ) 1 V Scope 1 Scope 2 Scope 4 Figure C. q).5 Scope 7 Ga i n2 Ga i n1 Kr Scope 5 r Fcn 1/KH2 tH2. in order to make the computation of all real quantities. the full-order synchonous generator model is connected to gas turbine.APPENDIX C.

1 4 . Iu (i)= (Iq (i)+ jay * Id (i))* e x p (jay * d elta (i)). g a s-tu rb in e D G . Gii Bii .. x d 1 ) % N etw o rk Interfa c e % S u b sta tio n . E d 1 (i)= u ((i-1 )* 3+ 2).0 ]). d elta (i)= u ((i-1 )*3 + 3 ). Ixi = Iyi .. Vxn Vyn Vx1 (C.APPENDIX C. d elta (i)= u ((i-1 )*3 + 1 ).......0 . Iu (i)= (E q 1 (i)+ jay * E d 1 (i))* ex p (jay *d elta (i))/ (jay * x d 1 (i)).0 .. fo r i= 2 :n _ v Iq (i)= u ((i-1 )* 3 + 2 )/ sq rt(3 )...T _ sw . Ixn Iyn Ix1 G11 B11 Gi1 Bi1 Gn1 B11 −G11 Bi1 −Gi1 Bn1 .n _ v . G1i B1i . i= 1 .0 . G1n B1n Gin Bin Gnn Bnn Bn1 −Gn1 Bni −Gni −G1n Bin −Gin Bnn −Gnn B1n Vy1 .Y _ fa u lt. d ie se l-en g in e D G % Z h ix in M ia o % A u g . Id (i)= u ((i-1 )* 3 + 3 )/ sq rt(3 ).. n = len gth (Y ).2) A simple example is used in as the network function in Matlab is shown as below: fu n c tio n o u to u t= fu n _ test(u ./ (jay * x d 1 ) .1 )... E q 1 (i)= u ((i-1 )* 3 + 1 ). 20 0 1 jay = sq rt(-1 ). PROGRAM STRUCTURE OF SIMULATION 160 Iy1 . Gni B1i −G1i Bii −Gii Bni . end if (t<= T _ sw (1 ) |t>= T _ sw (2 )) Y 1= Y + d ia g ([1 . Iu = ze ros(n .% 1 / (j* x d 1) . Vxi Vyi . Y .. t = u (n _ v * 3 + 1 )..

4 Features This software package setup a platform to study the DG models.2 * n _ v . V m (i)= V (i)* e x p (-jay * d elta (i)).0 ]). I= Y * V .0. eigenvalues and eigensensitivity calculation. most of the linear system toolbox in MATLAB can be used by the linearized model. residue calculation.APPENDIX C. V = inv (Y 1 )* Iu .’.1 ). participation matrix. V d (i)= im ag (V m (i)). I= Y _ fa u lt* V . end fo r i= 1 :n _ v Im (i)= I(i)* ex p (-jay * d e lta (i)). the impacts of the DGs on the power distribution system. Id (i)= im a g (Im (i)).3 Analysis Tools The analysis tools are developed to evaluate the linearized system model and to design the dynamic stability controller to improve the oscillation damping of the system. V q (i)= rea l(V m (i)). Also. The features of the software package can be summarized: .0. C./(jay * x d 1 ). critical eigenvalues identiﬁcation. e lse Y 1= Y _ fa u lt+ d ia g ([1 .V d (i)].0 .:)= [V q (i). o u t(i. and to design the controller improve the dynamic stability of the system. C. output feedback optimal control design. The analysis tools consist of: nonlinear models’ linearization in Simulink. end o u to u t= resh a p e (o u t. PROGRAM STRUCTURE OF SIMULATION 161 V = inv (Y 1 )* Iu . Iq (i)= rea l(Im (i)).

. PROGRAM STRUCTURE OF SIMULATION 162 1) Each DG model is developed independently. 3) The Simulink allows all the basic functions of the Simulink such as the numerical calculation methodology. 4) Modular design allows the software is very convenient to be expanded. so it can be used to investigate the characteristics of each speciﬁc DG. This uniﬁed structure makes it easier to connect multi-DGs to the power distribution system. the measurement units.APPENDIX C. to include the newly emerged DGs. 2) The inputs and outputs of each DG model are uniﬁed as current (I) and voltage (V) respectively.

:).1 :2 ).:)= B 2 r0 (1 0 . C 2 r0 = [C 2 r0 (1 :2 .C 2 (:. B0 . A 2_ z1 (1 0 .:).:). and the parameter uncertainty ∆A = A1 − A0 .:)].:). B 2 r0 (1 0.C 2 . [n 1 .:)= B 2 r0 (4 .:)-A 2 _ z1 (3 .163 Appendix D Robust Analysis Program Given linear system (A0 .:)]. B 2 r0 (4 . C0 ). % % step 2 : ca lcu la te freq u en c y re sp o n se o f M 1 1 .:)]. B 0 = [B 2 r0 (1 :2 . A 0 = [A 2 r0 (:. A 2 r0 = [A 2 _ z1 (1 :2 .:). [m .:)].:)= A 2 _ z 1 (4 .m 1 ]= size (C 2 ). (A0 . B0 .n ]= siz e(A 2 ).:)-A 2 _ z1 (3 . [A 2 . C 2 r0 = [C 2 (:.:).B 2 r0 (4 :m .:)-B 2 r0 (3 .:)-B 2 r0 (3 .n ]= siz e(A 2 ). A 2 _ z1 (4 :m .D 2 ]= lin m o d (’m a in _ sy s_ w in d _ svc_ A 1 6 _ lin ea r’).4 :m 1 )]. The robust analysis program is as the following: % % lin e a riz th e n o n lin ea r m o d e l: ’m a in _ sy s_ w in d _ svc_ A 1 6 _ lin ea r’. C0 ) is obtained from the nonlinear model in Matlab/Simulink. C 0 = [C 2 r0 ((n 1 -3 ):(n 1 -1 ).4 :m )]. A 2_ z1 = A 2 .:)= A 2 _ z1 (1 0 . where A1 is the system coeﬃcient matrix when the dynamic load has an increase.:). C 2 r0 (4 :n 1 . B 2 r0 = B 2. [m .1 :2 ).B 2 .A 2 r0 (:. A 2 _ z1 (4 .

M 1 g = frsp (M . [B 0 E ]. [C 0 (1 . ROBUST ANALYSIS PROGRAM 164 G = p ck (A 0 . o m ).1 .F ]).k k 0 .-1 .0].1 ). M = sta rp (G .0 .0 .:). % % step 4 : fu n ctio n m u to o b ta in th e u p p e r a n d low e r b o u n d o f m u b n d s1 = m u (M 1 1 _ g .u n cb lk ).1 :3 .8 0 ).APPENDIX D.1 :3 ).2 .-1 . M 1 1 _ g = se l(M 1 g . . o m = lo g sp a c e(-2 . % % step 3 : d escrib e th e stru ctu re o f th e p ertu rb a tion s u n cb lk = [-1 .

S. and received his Master of Science degree in 1997. . Zhixin Miao worked as a research assistant in Distributed Generation projects in West Virginia University. student. and M. In 1994. From 1992 to 1994 and from 1997 to 1999. he entered the graduate school of Nanjing Automation Research Institute. P. In 1999. China. China) in 1988 and got his Bachelor of Science degree in 1992.165 VITA Zhixin Miao was born in October 1970 in Nanjing. P. Jiangsu Province. China.R. he worked as an electrical engineer in power system protection department of Nanjing Automation Research Institute.R. He entered Huazhong University of Science & Technology (Wuhan. Both his B. degrees are in electrical engineering.D. Ministry of Electric Power. He was married with Lingling Fan in 1997. he enrolled West Virginia University as a Ph.S.

Sign up to vote on this title

UsefulNot useful- EEE 1201 021 Doubly Fed Induction Wind Turbines PSCAD EMTDC
- PSSE Fault Intro Instructions
- Cost and Performance Baseline for Fossil Energy Plants Volume 1_ Bituminous Coal and Natural Gas to Electricity
- dg nigeria case study
- The Lowest Cost Electricity for a Poor Rural Village in India- Rural Grid or Off-grid Spv
- DISTRIBUTED GENERATION
- Economic Evaluation of Embedded Generation in Nigeria
- Economic Evaluation of Projects
- Diesel Electric Generating Plants
- Operating and Planning Electricity Grids with Variable Renewable Generation
- Power Generation Financial Modelling and Analysis a Practical Guide
- Distributed Generation
- Power Plant Capital Cost
- Electric Power Research Institute
- Electrical Design Guidelines
- lm gas turbine.pdf
- Micro Turbine Generator System
- CCPP
- 2011 Charchalis Diagnosing of Ship Propulsion Elements
- Hurto a PDVSA (3)
- Hurto a PDVSA
- Gas Turbines for Electrical Power Production
- 55836500-Pwps-Mobilepac-Brochure.pdf
- Pwps Mobilepac Brochure
- Combined Cycle Power Plants
- gas turbine technology
- NES 309 Requirements for Gas Turbines Category 2
- Ansaldo-Original_Service_Provider_Brochure
- mel712-29
- 6340