AN ADVANCED REACTIVE POWER MANAGEMENT SYSTEM FOR THE SEOUL METROPOLITAN POWER SYSTEM

Transcription

1 AN ADVANCED REACTIVE POWER MANAGEMENT SYSTEM FOR THE SEOUL METROPOLITAN POWER SYSTEM Scott G. Ghiocel 1, Sangwook Han 2, Byung-Hoon Chang 3, Yong-gu Ha 3, Byong-Jun Lee 2, Joe H. Chow 1, and Robert Entriken 4 1 Rensselaer Polytechnic Institute (RPI) 2 Korea University (KU) 3 Korea Electric Power Company (KEPCO) 4 Electric Power Research Institute (EPRI)

2 AGENDA Overview Features Stability margin calculation Optimization Real-time dispatch Operator training Architecture Components Software 2

3 OVERVIEW Korean power system The Seoul metropolitan area is supplied by distant coastal generation, which creates potential voltage stability issues. Over the last few years, KEPCO has installed 2 SVCs and a STATCOM in the Seoul metropolitan area. Goals Coordinate the three FACTS controllers with local switched shunt capacitors and reactors Maintain adequate voltage stability margin Reduce shunt switching Maintain steady-state voltages 3

4 KOREAN POWER SYSTEM Seoul metropolitan area 4

5 MULTIPLE FACTS CONTROL SYSTEM (MFCS) COMPONENTS Multiple FACTS Controller (MFC) calculates FACTS controller settings for voltage stability. Reactive Power Manager coordinates FACTS controllers and other reactive power resources. Implements settings from the MFC algorithm. Minimizes shunt switching through optimization. Background daemon program monitors EMS output for new state estimator solution MySQL database data storage InTouch HMI user interface 5

6 SYSTEM ARCHITECTURE 6

7 ALGORITHM 1. New state estimator solution arrives. 2. Reactive power manager starts the MFC algorithm. 3. While the MFC algorithm is working, the manager calculates the VQ sensitivities and loads the SE data into the database. 4. After the MFC is complete, the manager retrieves the MFC results and loads them into the database. 5. The manager uses the MFC results to calculate the reactive power dispatch, and loads the results into the database. 7

8 COMPONENT: MFC ALGORITHM 8

9 MFC ALGORITHM I PURPOSE: Determine the necessary dynamic Q reserve to withstand major contingencies. INPUTS: State estimator solution (RAW file) FACTS controller parameters Contingencies Algorithm parameters OUTPUTS: FACTS controller settings: V references, Q reserves FV and YV curve data Inductive 0 Uses continuation methods to evaluate stability margins. Control the operating point Capacitive Secured margin 9

10 MFC ALGORITHM II YV Analysis FV Analysis 1: Pre-Contingency 2: Post-Contingency Bus Voltage [pu] λ max Bus Voltage [pu] Pre-Contingency Operating Point Post-Contingency Operating Point Flow Margin 0 Parameter λ Flows (MW) 1: Stable 2: Marginal 3: Unstable 10

11 MFC ALGORITHM III State estimator data Stable? Stable? Emergency Check Alarm Check Normal conditions YV Analysis Stable? No Yes Calculate Q reserves according to sensitivity analysis result YV Analysis after adjusting Q reserves No Stable? Yes Determine the Q reserves and Vref of each substation for stability FV Analysis Stable? No Yes Calculate Q reserves according to sensitivity analysis result FV Analysis after Adjusting Q reserves No Stable? Yes Determine the Q reserves and Vref of each substation for stability Individual FACTS controllers OPF module Objective function : P Loss minimization NIPM algorithm Determine Vrefs of each substation for Loss minimazation (need not Q reserves) 11

12 COMPONENT: REACTIVE POWER MANAGER 12

13 REACTIVE POWER MANAGER I PURPOSE: Coordinate dispatch of FACTS controllers and other reactive power resources (optimization). INPUTS: State Estimator solution (PSS/e RAW file) FACTS controller setpoints (V references, Q reserves) System forecast data Algorithm parameters OUTPUTS: New dispatch for FACTS controllers and switched shunts VQ sensitivities 13

14 SENSITIVITY-BASED REACTIVE POWER DISPATCH I The system VQ sensitivities (load lines) can be found from the load flow Jacobian by setting P = 0 from which we obtain the reduced Jacobian (JR) The reduced Jacobian matrix provides a linearized relationship between V and Q 14

15 SENSITIVITY-BASED REACTIVE POWER DISPATCH II Voltage V i Q k V op 2 V i Q k sensitivity of voltage magnitude at bus i due to reactive power insertion at bus k ΔV cap After insertion V op 1 V Vcap Qcap Qsh -Q capacitive ΔQ cap Before insertion Q inductive 15

16 SENSITIVITY-BASED REACTIVE POWER DISPATCH III After shunt capacitor Voltage V unc 2 V op 2 Droop Line Before shunt capacitor ΔV cap V ref V op 1 V unc 1 -Q capacitive Q inductive 16

17 SENSITIVITY-BASED REACTIVE POWER DISPATCH IV Comparison of Loadflow Results and Prediction using VQ Sens. dvlf dv 0.01 V (p.u.) Bus # 17

18 REACTIVE POWER MANAGER II Optimization: Maintain voltage stability (contingency survival) Implemented as constraints on the optimization Constraints on FACTS devices provided by the MFC algorithm Minimizing shunt switching action and device parameter changes Incorporated into the objective function as a cost per switching event Maintaining the desired voltage profile Incorporated into the objective function for specified pilot buses Voltage deviation from desired reference value 18

19 REACTIVE POWER MANAGER III Formulation: Linear program with continuous and discrete values (mixed integer linear program) Continuous values: FACTS Q setpoints and/or V references Discrete values: Shunt switching logic Constraints: Device limits Bus voltage limits Voltage stability limits (on FACTS controllers) Objective function: Number of switchings Voltage deviation at pilot buses FACTS deviation from setpoints 19

20 REACTIVE POWER MANAGER IV VQ sensitivities are used to predict the voltage effects due to reactive power injection, i.e., J Q V 1 R where J Jacobian. 1 R Bus voltage constraints is the inverse of the reduced load flow 1 V V J Q V V V LOW i.e., the new dispatch must adjust the new voltage to be within the specified bus voltage limits. Voltage stability constraints 0 R HIGH 0 Ensure adequate FACTS dynamic reserve: Q FACTS Q set 20

21 REACTIVE POWER MANAGER V OBJECTIVE FUNCTION: Terms c c c V V c Q Q set pilot 4 set FACTS 1. Number of shunt switchings 2. Number of FACTS V ref changes 3. Voltage deviation at pilot nodes 4. FACTS Q output deviation from setpoint values Ratio of weights in the objective determines priorities in the optimization. 21

22 REACTIVE POWER MANAGER VI System forecast: To optimize the shunt switching over longer periods of time (several hours ~ few days), past and future data are very useful. Future data (load forecasts) allows us to make smart decisions about switching shunts. Example: Keep a capacitor on in the morning instead of turning it off, in anticipation of a future load increase. Past data allows us to keep track of switching events, and can be used as an alternative for prediction if forecast data is not available. 22

23 DATABASE (MYSQL) PURPOSE: Stores network and dispatch data for MFCS. Enables communication between HMI and other components. STORED DATA: State Estimator solutions (new and old) MFC results FACTS setpoints FV/YV curve data MFCS dispatch results (new and old) Input from InTouch HMI Algorithm parameters Contingencies Note: Database should be implemented on a separate server from the rest of the system. 23

24 COMPONENT: MFCS HMI 24

25 MFCS HMI I PURPOSE: Displays MFCS data and allows the operator to change contingencies and algorithm parameters. DISPLAY SCREENS: One-line diagrams (present, past and future values) FACTS operator screens (including droop and load line) Device status (switched shunts, tap changers, etc.) Contingency analysis results MFC results MFCS configuration Contingencies Algorithm parameters (MFC and dispatch settings) 25

26 MFCS HMI III 26

27 MFCS HMI IV 27

28 MFCS HMI V 28

29 MFCS HMI VI 29

30 MFCS HMI VII 30

31 MFCS HMI VIII 31

32 OPERATOR TRAINING SIMULATOR I The MFCS can also be used in offline mode for operator training. Operators select from a list of cases (RAW files), and apply a contingency if desired. The MFCS runs the selected case, applying a contingency directly to the RAW file (if requested), and outputs the results to the InTouch HMI. MFCS can use any RAW file and contingency combination for offline simulation and training. 32

33 OPERATOR TRAINING SIMULATOR II Cases to consider: Line outages Daily variation Seasonal variation Contingencies: Single-line Double-line Interface line outages Metro area generator outages 33

34 IMPLEMENTATION Communication Links: SE data to MFCS: PSS/e version 30 RAW file Manager to/from MFC algorithm: ASCII text files Manager to/from database: MySQL client over TCP/IP MATLAB MySQL client over TCP/IP HMI to/from database: ODBC over TCP/IP (within control center firewall) MFCS to remote substations: TCP/IP over secured lines Software: MFC Algorithm: C++, FORTRAN Reactive Power Manager: MATLAB, Perl, MySQL, C++, C Database: MySQL HMI: WonderWare InTouch 34

35 THANK YOU FOR YOUR ATTENTION! 35