Hybrid Time Domain Simulation: Application to Fault Induced Delayed Voltage Recovery (FIDVR)

Transcription

1 Hybrid Time Domain Simulation: Application to Fault Induced Delayed Voltage Recovery (FIDVR) Vijay Vittal Arizona State University PSERC Webinar Jan. 2, 25

2 Presentation background The work presented in this webinar is a part of the work being done in PSERC project S-58 at ASU The PIs on this project are V. Vittal and S. Meliopoulos The graduate student at ASU is Qiuhua Huang and the collaborator is Dr. John Undrill 2

3 Outline. Introduction 2. Part -: OpenHybridSim: A new hybrid simulation platform 3. Part -2: Application to FIDVR study 4. Conclusions 3

4 Introduction: Background An increasing demand for simulating a portion of a system in detail with a very small time step and capturing point on wave detail while preserving the effects on the rest of the system Power electronic converter based renewable energy integration HV AC/DC systems Residential air conditioners, variable frequency drive(vfd) and other power electronics based loads Representing other needed detail at the distribution level PV Farm VSC-HVDC air conditioner 4

5 Introduction: Hybrid simulation Involves electromagnetic transient (EMT)- electro-mechanical transient stability(ts) hybrid simulation Instantaneous variable model EMT simulator Region of interest Boundary Remainder of the system Sequence, phasor model TS simulator Detailed System modeled in 3-phase Composite Load Model PV Converter Boundary bus i I abc i Z i Zik V k abc Boundary bus k Thévenin equivalent Z k V i abc + - V k abc + - Current injections I i I k Boundary bus i Boundary bus k External network Detailed system modeled in an EMT Simulator External system modeled in a TS Simulator 5

6 Introduction: Hybrid simulation Key requirements for a successful hybrid simulation platform Architecture : embedded/decoupled Interaction: communication + protocol Equivalent models of both detailed and external system Preparation and initialization of both detailed and external system Synopsis of literature review of EMT-TS hybrid simulation Lab research or proof-of-concept type The architecture design is not flexible and the simulators are limited to run on one computer Targeted mainly for three-phase balanced applications No publically available tool for hybrid simulation 6

7 Time scale of transient phenomena 7

8 OpenHybridSim: A new hybrid simulation tool Overall design Loosely decoupled architecture Socket communication InterPSS, an open source power system simulator A generic interface to an EMT simulator, e.g., PSCAD, Maltab/SimPowerSystems EMT Simulator Socket Component Socket Communication I EMT () t V abc Tt () InterPSS Core Engine Network Equivalent Helper Socket Server HybridSim Manager 8

9 Socket based communication framework TCP/IP socket based communication framework Enables decoupling of the EMT and TS simulators Supports application environment - Single computer - Local area network(lan) - Internet Socket components in PSCAD and Matlab/SimPowerSystems are developed for interfacing Socket component in PSCAD 9

10 External system equivalent for EMT simulation Supports both -phase and 3-phase equivalents -phase: mainly for three-phase balanced applications 3-phase: any type of fault within the detailed system Facilitates simulation of unsymmetrical faults 3-phase Thévenin equivalent of external system based on a 3-sequence network model The given base case is modeled using 3-sequences Details are provided on the next slide

11 Three-phase Thévenin equivalents of the external system Step : Calculate 3-sequence Norton equivalents Step 2: 3-sequence to 3-phase transformation for each boundary bus I abc Ni = SI Ni y = Sy S abc i i where S is the transformation matrix, is the bus primitive admittance Detailed system bus i abc z ik z iac z iab z ibc z ia z ib z ic + _ +_ +_ a V Ti b V Ti c V Ti Step 3: -phase Norton to Thévenin for each boundary bus p p V = I / y z = / Ti Ni ip where p stands for one of the three phases ip y ip bus k Single-phase line z kac z kab z kbc z ka z kb z kc Three-phase line +_ + _ +_ a V Tk b V Tk c V Tk

12 Detailed system represented by sequence current sources in TS simulation Need to extract three-sequence, fundamental frequency currents from instantaneous, waveform values Use well-established FFT algorithm PSCAD provides an FFT component Directly used in the network solution (I=YV) in TS simulation FFT Component in PSCAD 2

13 Three-sequence based TS simulation for simulating unsymmetrical faults () I EMT () t (2) I EMT () t () I EMT () t Positive-sequence network solution and integration step () xt () = f( xt (), y() t ) xt ( + T) = xt () + xt () Negative- and zerosequence network solver = g ( y, I ) (2) (2) 2 ( t + T ) EMT () t = g ( y, I ) () () ( t + T ) EMT () t T = g( xt ( + ), y, I ) () () T ( t + T ) EMT () t () V TS ( t + T ) (2) V TS ( t + T ) () V TS ( t + T ) 3

14 (a) series (b) parallel (c) combined Parallel Protocol... The interaction protocols between two simulators T T+ΔT... T+2ΔT T+3ΔT T T+ΔT T+2ΔT T+3ΔT T T+ΔT T+2ΔT T+3ΔT Fault on Series Protocol TS EMT TS EMT Fault cleared T+4ΔT T+5ΔT Parallel Protocol T+6ΔT TS EMT 4

15 Implementation with the series type protocol The series protocol Use the updated equivalent data for simulation with each simulator Performance issue: One simulator has to remain idle when the other is running the simulation xt () V TS () t I EMT () t socket server () I EMT () t Step () Step (2) Positive-sequence network solution and integration step () xt () = f( xt (), y() t ) xt ( + T) = xt () + xt () Negative- and zero- T = g( xt ( + ), y, I ) () () T ( t + T ) EMT () t V () TS ( t + T ) (2) I (2) EMT () t sequence network solver V TS ( t + T ) (2) (2) = g2 ( y( t + T ), IEMT () t ) () () I V EMT () t () () TS ( t + T ) = g ( y( t + T ), IEMT () t ) InterPSS V TS( t + T) I EMT () t Calculate 3-phase Thévenin equivalent Step (3) V abc Three-sequence TS simulation Tt ( + T ) V abc Tt ( + T ) socket server xt ( + T) V TS ( t + T ) I EMT () t t socket client EMT step EMT step 2 EMT step 3 Step (5) EMT Step (4).. EMT step N- EMT step N I EMT ( t+ T ) t + T 5

16 Implementation with the parallel type protocol The parallel interaction protocol Steps (d) and (e) run simultaneously May cause significant interfacing errors under fast transient conditions, as the Thévenin equivalent data is one-step delayed xt () V TS () t I EMT () t I EMT () t socket server socket client Step (a) V TS () t I EMT () t EMT step Calculate three-phase Thévenin equivalent EMT step 2 socket server... EMT step N- Threesequence TS simulation Step (b) Step (c) Step (d) V abc Tt () Step (e) EMT step N xt ( + T) V TS ( t + T ) I EMT ( t+ T ) t t+ T 6

17 Automatic protocol switching algorithm Automatically switch interaction protocol based on the system conditions, reflected by the maximum rate of change of sequence current injections I ( p) ( p) I I EMT (, i t) EMT (, i t T) RI = max( max ( )) / T EMT () t () i p (, 2, ) I EMT (, i t T ) I EMT () t EMT ( t T ) Maximum rate of change RI EMT time step T Y RI N EMT > ε? series N last step used series? Delay Logic of the protocol switching algorithm Y parallel 7

18 Automatic creation and initialization of the external network Base case I i Boundary bus i bus m Boundary Configuration Depth first search(dfs) to identify detailed system Detailed System I k V k bus k Boundary The remainder of the external system bus n I i I k bus i bus k V k bus m The remainder of the external system bus n ) Set buses and branches within the detailed system out of service 2) initialize the external system Power flow 8

19 Test system Testing of the developed platform G2 Bus2 8 kv 92MVA T2 X T =.625 Bus 8 23 kv Bus 7 +j35mva Bus 9 Bus 3 23 kv 23 kv 3.8 kv.85+j.72.9+j.8 G3 Bus 5 23 kv 25+j5MVA.32+j.6.+j.85 T X T =.576 G.7+j j.7 Bus kv Bus 6 23 kv 9+j3MVA T3 28MVA X T =.586 Bus 4 23 kv (case ) 247.5MVA The IEEE 9 Bus system Modeling of Bus 5 and the external network Thévenin equivalents in PSCAD 9

20 Case : The interaction protocols Case setting: - EMT time step = 5 us -TS time step = 5 ms Protocol control setting: -Threshold ε =.4 - Delay = 2 cycles A three-phase fault at. s and lasts for 4 cycles Plots (Top) Positive-sequence current injection at bus 5 into the external network (Bottom) Protocol switching signal current (ka) PSCAD EMT-TS(series) EMT-TS(parallel) EMT-TS(combined) (a) log(ri ) EMT(t) log(.4) protocol signal Time (s) (b) 2

21 G2 Bus2 8 kv 92MVA T2 X T =.625 Bus 8 23 kv Bus 7 +j35mva Bus 9 Bus 3 23 kv 23 kv 3.8 kv.85+j.72.9+j.8 G3 25+j5MVA Case 2: Performance of the developed platform under unsymmetrical faults Internal network: Bus 5 and branch Bus5-Bus7 Two-port three-phase Thévenin equivalent Detailed modeling of bus 5 using the WECC composite load model, with the -p air conditioner compressor motor represented by a EMTP type model developed by Yuan Liu, et al [] Bus 5 23 kv.32+j.6.+j.85 T X T =.576 G.7+j MVA.39+j.7 Bus kv Bus 6 23 kv 9+j3MVA T3 28MVA X T =.586 Bus 4 23 kv Single line to ground fault starting at 2. s 4 cycles 2.47 kv Equivalent feeder Composite Load Model Bus 5 23 kv 23 kv/69kv 5 MVA % 69/2.47 kv 5 MVA 8% 2.47 kv Composite Load Model 69kV 69/2.47 kv 5 MVA 8% 2.47 kv Composite Load Model Detailed modeling of Bus 5 [] Y. Liu, V. Vittal, J. Undrill, and J. H. Eto, "Transient Model of Air-Conditioner Compressor Single Phase Induction Motor," IEEE Transactions on Power Systems, vol. 28, pp , 23. 2

22 Case 2: Performance of the developed platform under unsymmetrical faults 22 PSCAD EMT-TS(parallel) EMT-TS(combined) protocol signal Voltage (kv) Phase A PSCAD EMT-TS(parallel) EMT-TS(combined) protocol signal Voltage (kv) Voltage (kv) Phase B Phase C Time (s) Three-phase voltage of Bus 5 Current (ka) Time (s) The phase A current injection at Bus 5 into the external network

23 Takeaways of part-. The combination of the decoupled architecture and socket-based communication facilitates both simulators to be run on either one computer or several computers to achieve more flexibility and a better performance 2. The proposed combined interaction protocol with the auto-switching feature helps improve the hybrid simulation efficiency while a good accuracy is guaranteed 3. Application of the multi-port three-phase Thévenin equivalent enables simulating unbalanced faults within the detailed system, without the constraint of phase balance at the boundary buses 23

24 Fault induced delayed voltage recovery (FIDVR) Fault distribution, subtransmission and transmission systems Delayed voltage recovery several to tens of seconds Root cause of FIDVR air conditioner (A/C) compressor motor stalling and prolonged tripping Voltage profile during a typical FIDVR event [2] [2] D. N. Kosterev, A. Meklin, J. Undrill, B. Lesieutre, W. Price, D. Chassin, et al., "Load modeling in power system studies: WECC progress update," in 28 IEEE Power and Energy Society General Meeting - Conversion and Delivery of Electrical Energy in the 2st Century, 28, pp

25 Fault induced delayed voltage recovery (FIDVR) Accurate FIDVR studies require Detailed A/C modeling Detailed network model down to distribution feeder level Limitations of the conventional positive-sequence TS simulator Distribution system configurations Response of single-phase devices when subjected to unsymmetrical faults FIDVR events are generally localized Detailed modeling can be limited to a small portion of a large power system FIDVR Area A large power system 25

26 Determination of the boundary of detailed system Use a simple yet generic test case to quantify the voltage dip threshold at a transmission bus causing A/C motor stalling Criterion: A bus is included in the internal system if a singlephase or three-phase fault at that bus causes a phase-to-neutral voltage at buses with a large percentage of A/C motor loads to drop below.75 pu. x T =.25 x T =.5 x T =. x T =.5 x T = kv.4 pu s Equivalent Step-down Equivalent source transformer feeder impedance Bus Bus 2 impedance Bus j.5 X T = 2.5 2%.3 + j.4 5/2.47 kv Composite Load model Voltage dip magnitude (pu) A/C load percentage (a) Voltage dip magnitude (pu) A/C load percentage (b) Voltage dip magnitude w.r.t. the A/C load percentage and the transformer impedance: (a) A/C power = 4.9 kw (b) A/C power = 6. kw 26

27 Buses Applied the boundary selection criterion to the WECC system The WECC system Transmission lines Generators Loads Buses with a large percentage of -Ф A/C motor load Bus 245 Bus sub sub 2438 sub 5 kv M M M Load Load sub sub NORTH sub EAST One-line diagram of the study region sub 5 kv Sub-system below 5 kv STATISTICS OF THE DETAILED SYSTEM MODEL Total number of buses Number of buses of different voltage levels Total Load kv 7 23 kv 37 6 kv 3 5 kv kv 8 <= 66kV 5.9 GW 27

28 Initialization of the detailed system with a large percentage of load as induction motors The built-in initialization function of PSCAD fails to initialize the detailed system A two-step initialization approach () The switch K is turned to the position, and the distribution systems and the CLMs are energized by the fixed voltage sources and initialized independently. (2) After the CLMs are successfully initialized, the switch K is turned to the position such that distribution systems are connected to the transmission system. Bus HV LV s K Distribution network Static equivalent load CLM CLM CLM NOTE: The magnitude and phase angle of the fixed voltage source, S, are set based on the power flow solution of the given base case CLM: composite load model 28

29 Benchmarking hybrid simulation against conventional transient stability Both use constant impedance load A single line to ground (SLG) fault is applied on the phase A of bus 245 at t =.2 s Voltage magnitude(pu) Bus 245 positive sequence voltage EMT-TS InterPSS Time (s) Machine reactive power (pu).5 Voltage magnitude(pu).95.9 Bus 248 positive sequence voltage EMT-TS InterPSS Time (s) Bus493 machine # reactive power EMT-TS InterPSS Time(s) 29

30 FIDVR event triggered by a SLG fault Detailed modeling of the region served by Bus 245 Bus kv 45 +j*75 MVA 45 +j*75 MVA 43 +j*96 MVA 43 +j*96 MVA.2536 pu 56 MVA 53.65/ kv.2536 pu 56 MVA 53.65/ kv.2536 pu 56 MVA 53.65/ kv.2536/.4/.52 pu 56 MVA Bus kv Bus kv 22 + j*. MVA Bus kv 46.8 MVAr.2 pu 25 MVA 5 /2.47 kv.2 pu 25 MVA 5 /2.47 kv M M M Equivalent Feeder f- f-2 f-3 f-4 f-5 f-6 f-7 f-8 Equivalent feeder Model Equivalent feeder Model Equivalent feeder Model Equivalent feeder Model Equivalent feeder Model Equivalent feeder Model Equivalent feeder Model Equivalent feeder Model 22+j*(-8) MVA 22+j*(-8) MVA 22+j*(-8) MVA 22+j*(-8) MVA 22+j*(-8) MVA 22+j*(-8) MVA 22+j*(-8) MVA 22+j*(-8) MVA Feeder modeled in two sections Z 4 feeder M A B C 3 Z 4 feeder M A/C 3

31 FIDVR event triggered by a SLG fault (cont ) 7% -Ф A/C motor, 5% 3-Ф induction motor, 5% constant impedance Phase A and C were directly affected by the SLG fault at Bus 245 A/C motor stalling propagated to nonaffected phases (phase B in this case) Voltages of all three phases were depressed after.95 s Terminal voltage of the A/C motors at the /4 length point phase A.5 phase B phase C Speeds of the A/C motors at the /4 length point.5.5 phase A phase B phase C Speeds of the A/C motors at the end of the feeder phase A phase B phase C Sequence voltages of the 5 kv bus# 246 Positive.5 Negative Zero Sequence voltages of the 5 kv bus# 245 Positive.5 Negative Zero

32 Effects of load composition on FIDVR Case Load composition # 75% -Ф A/C motor, 25% constant impedance 2 75% -Ф A/C motor, % 3-Φ NEMA type B induction motor, 5% constant impedance 3 7% -Ф A/C motor, 5% 3-Ф NEMA type B induction motor, 5% constant impedance 4 6% -Ф A/C motor, 5% 3-Ф NEMA type B induction motor, 25% constant impedance 5 5% -Ф A/C motor, 25% 3-Ф NEMA type B induction motor, 25% constant impedance Propagation of A/C motor stalling to unfaulted phase is consistent across a substantial range of load compositions The impacts of SLG faults close to certain regions of the system with high A/C penetration could be more severe than perceived Speed (pu) Speed (pu) Speed (pu) Speed (pu) Speed (pu).5.5 phase A Case phase B phase C Case Case Case Case Case Time (s) 32

33 Effects of point-on-wave (POW) on FIDVR POW effects on the occurrence and evolution of FIDVR are apparent, based on the differences in the response of the A/C motors of bus 246 The POW when the fault occurred should be considered for detailed FIDVR study Speed(pu) Voltage(pu) Positive Seq.Voltage(pu) Phase A Phase B Phase C bus 245 bus t(s) Case 5A, POW = 45 deg (a) (b) (c) Speed(pu) Voltage(pu) Positive Seq.Voltage(pu) bus 245 bus t(s) Case 5, POW = deg (a2) (b2) (c2) 33

34 Takeaways of part-2. The study shows that a normally cleared, single line to ground fault at a 5 kv bus close to the A/C loads can lead to a FIDVR event. The event begins with A/Cs stalling on two directly-impacted phases, followed by A/C stalling propagating to the unimpacted phase. 2. Further, five study cases with quite different load compositions show similar A/C motor stalling results. 3. The POW when the fault occurs could have a significant impact on the response of the A/C compressor motors. 34

35 Conclusions. A new EMT-TS hybrid simulation tool is developed, which features ) decoupling architecture, 2) generic interfacing with an EMT simulator, 3) flexible switching of interaction protocol and 4) support of three-phase equivalent of external system. 2. The hybrid simulation tool has been applied to study the FIDVR phenomenon within a region of WECC system, certain aspects of the evolution of the phenomenon were uncovered for the first time. 3. The developed tool is not limited to FIDVR study, but also applicable to studies that require detailed models and simulation of a part of a large power system, while preserving the slow dynamics of the rest of the system 35

36 Thank you! 36