Implementation and Evaluation of a Labscale Synchrophasor Model and Applications

Transcription

1 Implementation and Evaluation of a Labscale Synchrophasor Model and Applications Imran Sharieff, M Prasad, K N Satish and Ranjana Sodhi Dept of Electrical Engineering, IIT Ropar, Pb 11 India {imrans, prasadm, satishkn, rsodhi}@iitrpracin Abstract This paper presents a preliminary implementation of a Labscale Synchrophasor model, where one SEL- (Satellite Synchronized Clock) and SEL-51 relay is installed and properly configured The synchrophasor data is metered in realtime and displayed graphically For the purpose of analysis and developing the backend applications, the Phasor Measurement Unit (PMU) data is archived using openpdc An important backend power system application is then implemented which utilizes the archived synchrophasor data The application implements an algorithm to screen the moving window of PMU data for event detection The algorithm uses three different methods to screen for possible events viz, Min-Max method, Fast Fourier Transform method and Pencil-Matrix method Each method is applied to a moving window of PMU data The event detection algorithm is tested on the PMU data obtained from the Real Time Digital Simulator (RTDS) Index Terms Phasor Measurement Unit (PMU), Synchrophasor measurements, Event detection, Pencil Matrix method I INTRODUCTION /1/$31 c 1 IEEE Due to the increasing development of power networks, their control systems and protection requirements are becoming complex In recent years, the implementation of Phasor Measurement Unit (PMU) based synchrophasor technology has verified that it brings deep-seated advantages for the power system monitoring, control and protection [1] This paper focuses mainly on the power system monitoring applications that can be developed utilizing the Synchrophasor data Since PMU is the building block of the synchrophasor technology, it is worthwhile to have a hands-on experience with the device To this end, a Lab-Scale Synchrophasor Model is developed at IIT Ropar This includes installation of SEL- Satellite Synchronized Clock [] and SEL-51 relay [3] and their configuration to meter and graphically display the synchrophasors This is followed by archiving the data for further processing The PMU data is archived using openpdc [] Having developed a Lab-Scale Synchrophasor Model and archived the real-time data, an algorithm is implemented to screen a small, moving window of PMU data for event detection The algorithm can help the back-end operator since he does not need to analyze the entire voluminous data at once, but instead analyze a window of PMU data The efficacy of the implemented event detection algorithm is demonstrated on test cases, simulated in the RTDS The rest of the paper is organized in four sections II describes the various steps involved in setting up the synchrophasor model at a labscale III presents the event detection algorithm and its test results The main conclusions are drawn in IV of the paper II DEVELOPING A LABSCALE SYNCHROPHASOR MODEL A typical architecture of a Synchrophasor Model comprises of the Phasor Measurement Units (PMUs), installed at various locations in the power system These PMUs measure voltage phasors, current phasors and frequency, generally using the Discrete Fourier Transform (DFT) [1] PMUs use the IEEE C37118 data format for communication with the central monitoring station, called the Phasor Data Concentrator (PDC) The PDCs can be hardware based as well as software based data concentrators The typical functions of the PDC are to collect data from several PMUs, reject bad data, align the time stamps and create a coherent record of simultaneously measured data from wider part of the power system and utilize these measurements for further applications To mimic a typical synchrophasor model, following components are used with the configuration settings as described in the subsequent subsections A Basic Components 1) SEL- Satellite-Synchronized Clock []: This Satellite Synchronized Clock provides demodulated IRIG-B time code with time-quality values specified by IEEE C37118 (Standards for Synchrophasors for Power Systems) The SEL- includes four programmable IRIG-B outputs, capable of providing ± 1 ns accurate time signals to multiple devices One SEL- Satellite-Synchronized Clock can provide time inputs to four SEL Synchrophasor-equipped devices in a station ) SEL-51-5 Relay [3]: The SEL-51 relay is a distribution relay featuring auto-reclosing with synchronism check, circuit breaker monitoring and circuit breaker failure protection The SEL-51 features extensive metering and data recording including high-resolution data capture and reporting The relay provides extensive communications interfaces from standard SEL, ASCII and enhanced MIRRORED BITS communications protocols to Ethernet connectivity, with the optional Ethernet card With the Ethernet card, the latest industry communications tools, including Telnet, FTP, IEC 6185, and DNP3 (Serial and LAN/WAN) protocols can be employed The acselerator Quickset software program is included with the SEL-51 acselerator Quickset assists in

2 setting, controlling, and acquiring data from the relays, both locally and remotely The SEL-51 supports IEEE C Standard for Synchrophasors for power systems SEL51 configured as a PMU: The SEL-51 is useful to apply over-current protection for primary and backup, and three-pole tripping schemes on sub-transmission lines Nonetheless, it can also function as a PMU as it provides bestin-class protective relay functions C37118 message format allows up to 1 current and 8 voltage synchronized measurements, up to 6 messages per second (on a 6 Hz nominal power system) Five unique data streams, three choices of filter response, settable angle correction, and a choice of numeric representation makes the data usable for a variety of synchrophasor applications 3) acselerator QuickSet Software: Knowing how the power system functions and managing its information are critical to any operation acselerator QuickSet Software enables quick and easy configuration of SEL-51 for power system protection, control, metering, and monitoring Positivesequence phasor data from the PMU can be collected at an appropriate central site using a data concentrator or can be exchanged between local units for protection/control applications In this work, the communication protocols used are serial communications ie to send byte by byte or Standard Internet Protocol (IP) which can be carried over Ethernet There are four message types: data, configuration, header and command [3] Data messages give information about the measurements made by a PMU, configuration is a machine-readable message describing the data and providing calibration factors and header is human-readable descriptive information sent from the PMU but provided by the user Command messages are machine-readable codes, sent to the PMU for control or configuration ) PMU Connection tester: This application is used to view and archive the live stream of data in IEEE C37118 data format obtained from the PMU The real-time data consists of the time stamps, frequency, phase and magnitude of voltage and currents The real-time data thus obtained can further be utilized for real-time applications [5] B Configuring the Synchrophasor Model As discussed earlier, Synchrophasor measurement refers to the concept of providing measurements taken on a synchronized schedule in multiple locations A high-accuracy clock, commonly a Global Positioning System (GPS) receiver such as the SEL- Satellite-Synchronized Clock, makes synchrophasor measurement possible The availability of an accurate time reference over a large geographic area allows SEL-51 Relays, configured as PMU, to synchronize the gathering of power system data acselerator Quickset serves as the interface between the PC and SEL-51 relay to configure the device with proper settings Firstly, to meter and monitor the time-stamped test system data, the SEL-51 is configured with various settings as listed in Table I-II Global Settings: For the transmission of metered data to the PC, the device offers two means: one is through Serial Port and the other is through Ethernet Port [3] TABLE I GLOBAL SETTINGS FOR CONFIGURING THE PMU Setting Setting Prompt Default EPMU Synchronized Phasor Meas (Y,N) N MFRMT Message format (C37118,FM) C37118 MRATE Msg/Sec (1,,,5,1,1,15,,3,6) PMAPP PMU Application (F,N,1) N NUMPHDC Number Of Data Configurations(1-5) 1 PMSTN Station Nmae(16 characters) STATION A PMID PMU Hardware ID (1-6553) 1 Serial Port Settings: The port settings are used for configuring synchrophasor data transmission over a serial port [3] TABLE II SERIAL PORT 1,, 3, F SETTINGS FOR SYNCHROPHASORS Setting Description Default PROTOCOL (SEL, DNP, MBA, MBB, RTD, PMU) SEL SPEED Data Speed (3576) 96 STOPBIT Stop Bits (1, ) 1 RTSCTS Enable Hardware Handshaking (Y, N) N FASTOP Enable Fast Operate Messages (Y, N) N PMU MODE (CLIENTA, CLIENTB, SERVER) SERVER C Testing the Setup Figure 1 shows the setup of the SEL- device, where SEL- is powered by a 1V AC source and is having a GPS signal receiver, installed on the terrace, making sure that it has an unobstructed line of sight because four or more GPS satellites are required to provide location and time information The device includes four programmable IRIG-B outputs capable of providing ±1 ns accurate time signals to four multiple devices In the lab setup, only one of these four outputs (out-) is used, giving IRIG-B signal to SEL-51 Fig 1 SEL- Setup The preliminary lab setup of synchrophasor model is shown in Fig One of the IRIG-B output of SEL- is fed to the SEL-51 device for time synchronized measurements The

3 Fig SEL 51 Lab Setup V, 3-phase supply line is fed into a 3-phase autotransformer and the output voltage of 9V (rated input voltage for SEL- 51) is connected to one of the voltage channels of the PMU In the preliminary tests, the circuit is terminated by a starconnected balanced load with phase current of 7 A [6] In subsequent rigorous testing, the relay shall be installed at the 11 KV substation at the transit campus of IIT Ropar through current and potential transformers D Archiving the Synchrophasor Data Power system data, metered in real-time, can be displayed graphically and the data can be archived to develop backend power system applications The openpdc, administered by the Grid Protection Alliance (GPA), is a complete Phasor Data Concentrator software system It is designed to process streaming time-series data in real-time Measured data gathered with GPS-time from SEL-51 is time-sorted and archived in the PC The PMU Connection Tester is also used to view a live stream of data in one of the supported protocols in order to test that the stream is being received With the help of openpdc, the synchrophasor data is exported in MySQL [] Figure 3 shows the instantaneous metering values of the lab-setup Figure 3 shows the current phasors, voltage phasors, (both phase-phase and phase-neutral) and the real time-frequency The time stamp on the right top shows the small time-stamp The graphical representation of the phasors is shown in the Fig III IMPLEMENTING BACK-END APPLICATION FOR SYNCHROPHASOR DATA PROCESSING In II, the capabilities of PMU are examined by developing a Lab-scale Synchrophasor model, and metering the phasors and archiving the time-stamped data Having archived the PMU data, various back-end monitoring, protection and control applications can be developed which would utilize the PMU data In this section, a preliminary yet significant Fig 3 Fig Instantaneous metering values Phase Components monitoring application is implemented ie Event Detection using screening methods A Event Detection using Screening Methods Since synchrophasor measurements are captured from geographically dispersed PMUs, it is possible to identify system

4 trends and behavior such using eg power system frequency and voltage phase angle responses to an event The term event in this paper is used to describe transients due to large disturbances on the power system like short circuits and sudden changes like large and sudden loss of generation or load PMUs generate high precision data typically at a rate of 3 observations per second, resulting in large volumes of phasor data In larger wide-area monitoring systems containing a large number of PMUs, the amount of data generated can be difficult to handle To improve the situational awareness of the grid, a method is needed which would scan the voluminous phasor data in small batches and identify the possible event in moving window By analysing a phasor data window (say of past 1sec), the state of the system in next window can be estimated by effective processing A literature review reveals that there exist different spectral and parametric methods to detect different types of events by processing PMU signals [7] Three of such commonly used methods are implemented to detect the transient events in the system viz, Min-max method, Fast Fourier transform (FFT) and Matrix-Pencil method FFT is a spectral method, whereas matrix-pencil is parametric These methods are discussed as follows 1) Min-Max Method: This is the most simple and straightforward method to screen for an event in a window of PMU data of the PD signal In this method, the difference between maximum and minimum values within the data window is calculated The difference for each data window for the entire hour is saved This method is sensitive because it also detects gradual changes in the signal above the 1-second data window that are not caused by power system events Possible events are detected by calculating the average and standard deviation for all the data window differences for the entire hour of PMU data The differences that exceed three times the standard deviation are marked as possible events ) Fast Fourier Transform: The Fast Fourier Transform (FFT) method is based on using the maximum magnitude of the spectral content of PMU signals to detect possible events Before this method is applied, the data in the window is differentiated to remove DC offset in the data to facilitate the detection of events in the low-frequency range below 1 Hz Differentiating the data increases the noise in the data; however, it is found that the magnitude of frequencies associated with events in the data tend to be very high Therefore, increased noise is not a concern The FFT method is applied to each 1-second data window of the PMU data The discrete Fourier transform is used to find the strongest frequency component within the data window The average and standard deviation are calculated for all saved peak magnitudes Any saved magnitudes above three standard deviations are tagged as possible events Three standard deviations window was selected based on examples of detected events from the synchrophasor network data 3) Matrix Pencil: The matrix-pencil method is a parametric method that estimates parameters to fit the PMU signal In general, the signal model of the observed time-response can be formulated as, y(t) = x(t)+n(t) M R i exp(s i t)+n(t); i=1 t T (1) where y(t) = observed time response n(t) = noise in the system x(t) = signal R i = residues or complex amplitudes s i = α i + jw i, α i =damping factors, ω i = angular frequencies (ω i = πf i ) After sampling, the time variable, t, is replaced by kt s, where T s, is the sampling period The sequence can be rewritten as, and y(kt s ) = x(kt s )+n(kt s ) M R i zi k +n(kt s ); () i=1 z i = e sits = e ( αi+jωi)ts ; k = to N 1, i = 1,,M (3) Matrix Pencil is a linear technique to find the best estimates of M, R i s, and Z i s from y(kt) When two functions defined on a common interval, with a scalar parameter λ f(t,λ) = g(t) = λh(t) () f(t,λ) is called a pencil of functions g(t) and h(t), parameterized by λ In this implementation, one forms the data matrix [Y] from the noise-contaminated data y(t) by combining [Y 1 ] and [Y ] as y() y(1) y(l) y(1) y() y(l+1) [Y] = (5) y(n L 1) y(n L) y(n 1) The size of Y matrix is (N L) (L+1) For efficient noise filtering, the parameter L is chosen between N/3 to N/ For these values of L, the variance in parameters z i, due to noise, has been found to be minimum All the N data samples are utilized, even thoughlmay be considerably less thann Next, a singular value decomposition of the matrix [Y] is carried out as [Y] = [U][ ][V] H (6) Here, [U] and [V] are unitary matrices, composed of eigenvectors of [Y][Y] H and [Y] H [Y], respectively, and [ ] is a diagonal matrix containing the singular values of [Y], ie [U] H [Y][V] = [ ] (7) Next, the filtered matrix [V ] is constructed so that it contains only M dominant right singular vectors of [V] [V ] = [v 1,v,,v M ] (8)

5 The right-singular vectors from M +1 to L, corresponding to the small singular values, are discarded Therefore, [Y 1 ] = [U][ [Y ] = [U][ ][V1 ] H (9) ][V ] H (1) where [V 1 ] is obtained from [V ] with the last row of [V ] deleted; [V ] is obtained by removing the first row of[v ]; and [ ] is obtained from the M columns of [ ] corresponding to the M dominant singular values It can be shown that, for the noiseless case, the eigenvalues of the following matrix {[Y ] λ[y 1 ]} L M {[Y 1 ] + [Y ] λ[i]} M M (11) are equivalent to the eigenvalues of the following matrix {[V ] H λ[v 1 ] H } {[V 1 ] H } + {[V ] H } + λ[i] (1) This methodology of solving for z i s provides minimum variance in this estimate of z i s in the presence of noise Typically, up to -5 db of signal to noise ratio (SNR) can be handled adequately by this technique Once M and z i s are known, the residues, Ri, are solved for the following least squares problem: y() y(1) y(n 1) = z 1 z z M N 1 N 1 N 1 z 1 z z M B Test Results of Event Detection Algorithms R 1 R R M (13) This is to be noted that the labscale synchrophasor model developed in II does not have any test system where events can be created practically Therefore, to test the event detection algorithms, a benchmark system for HVDC is simulated in Real Time Digital Simulator (RTDS) at IIT Kanpur [8] To test the screening algorithm, a line to ground fault is simulated at the inverter side of the HVDC line and the relative phase angle difference (PD) of PMUs is archived for analysis Figure 5 shows the voltage magnitude and the PD data of the system simulated in RTDS Relative Phase Angle Difference between two separate PMU locations (PD) is used to screen an event in the power system To screen for events within each hourly PMU file, an overlapping, moving window is applied to the PMU data The window size is 1 seconds long and overlaps half of the previously windowed data It is clear from Fig 5 that the fault occurred around sec and is cleared at around 3 sec The algorithm should also be able to accurately detect the fault The PMU data from the RTDS is obtained at the rate of 1 data per 5 micro second, that is,, data per second Since the simulation is run for 1 seconds, so a total of,, PD data is to be analyzed The entire PD data is traversed through a window of length sec, ie, having data points in each window and the consecutive windows overlap midway Thus, a total of Fig 5 Fig 6 fault Voltage Magnitude (KV) Phase angle difference (radians) Time (Sec) Voltage Magnitude data of the PMU located near the fault site Time(sec) Relative Phase Angle Difference (PD) data of the System during 1 windows are analyzed The three methods, as discussed in previous subsection, ie Min-Max, FFT and Matrix Pencil are applied on each window The peak of each window is obtained and their mean & standard deviation are calculated The window bearing a peak which is 3 standard-deviation away from the mean is suspected for possible event 1) Response to the Min-Max Method: Figure 7 shows the results of Min-Max event screening algorithm It is observed from Fig7 that the windows which start at 19 second, second and 1 second may contain an event Since the width of each window is second So, the windows 19s to 1s; s to s and 1s to 3s may possibly contain the event Thus, taking the overlap of two consecutive windows, the event can be said to occur between s to s This is actually correct as it is evident from the voltage magnitude and phase differences plots of Figs5-6 Peak Magnitude of data window Peak Magnitude Min Max method 3*sigma from mean Window start time (sec), window width=sec Fig 7 Min-Max Method results on Phase Angle Difference

6 ) Response to the FFT Method: Figure 8 shows the results of FFT method It is observed from Fig8 that the windows which start at 19 second and second may contain an event Taking the overlap of two windows, FFT method indicates that the event can be said to occur between to 1 second Peak magnitude in data window Peak magnitude in data window Peak Magnitude FFT method 3*sigma from mean Window start time (sec), window width=sec Fig Fig 9 FFT Method applied on Phase Angle Difference Peak Magnitude Matrix Pencil method 3*sigma from mean Window start time (sec), window width=sec Matrix-Pencil Method applied on Phase Angle Difference 3) Response to the Matrix-Pencil Method: In this case also, the results of Fig 9 suspect that the event would have occurred between to 1 second After all methods have been used to screen for possible events in the PMU data, the time stamps of data windows marked as containing possible events are compared If two or more methods detect a possible event in the same data window, that data window is marked as containing an event Peak Magnitude in Window Min Max FFT Matrix Pencil Window start time (sec), window width= sec IV CONCLUSIONS This paper has considered the development of a basic Labscale Synchrophasor Model The devices SEL-51-5 (a distribution relay) and SEL- (Satellite Synchronized Clock) are installed in the lab SEL-51 is configured to work as PMU using acselerator Quickset software and the setup is tested using a simple three-phase load in the lab The PMU data is then archived using PMU connection Tester software The paper then implements a backend application for synchrophasor data processing An algorithm for synchrophasor-data based event detection using standard screening-methods Min-Max, FFT, and Matrix-Pencil methods are commonly used by the utilities, are implemented to detect the system-transients and their individual as well as collective responses are presented It is observed that min-max method screens for large fluctuations in the PMU data since it does not depend on frequency content of data The performance of the implemented event screening methods is found to be satisfactory on the phasor data, obtained from two PMUs in the RTDS Some related topics which have not been covered in this present paper like performance of the labscale synchrophasor model under varied load scenarios, categorization of the events are to be covered in a future paper V ACKNOWLEDGEMENT Authors acknowledge the partial financial support provided by the Department of Science & Technology (DST), New Delhi, under IIT Ropar project no SR/FTP/ETA-5/1, and the Initial Research Initiation Seed Grant, IIT Ropar, to carry out this research work Authors also greatly acknowledge the help and support provided by the RTDS Lab personals at IIT Kanpur REFERENCES [1] MG Adamiak, D Novosel, B Kasztenny, V Madani, J Sykes, and AG Phadke, Wide area protection and control - today and tomorrow, in Transmission and Distribution Conference and Exhibition, 5/6 IEEE PES, pages 17, May 6 [] Schweitzer Engineering Laboratories Inc SEL- Satellite- Synchronized Clock Instruction Manual, 13 [3] Schweitzer Engineering Laboratories Inc SEL-51-5 Instruction Manual, 13 [] Grid Protection Alliance, Open Phasor Data Concentrator, [5] Grid Protection Alliance, PMU Connection Tester, [6] Schweitzer Engineering Laboratories Inc SEL Synchrophasors- A New View of the Power System, 8 [7] E Muljadi, A Allen, and S Santoso, Algorithm for screening phasor measurement unit data for power system events and categories and common characteristics for events seen in phasor measurement unit relative phase-angle differences and frequency signals, National Renewable Energy Laboratory, pages 8 16, 13 [8] Real Time Digital Simulator Facility, IIT Kanpur, web rtds/ Fig 1 Comparison results of all three methods It is seen that all the three methods (in Fig 1) commonly suspect for the event in the windows 19s to 1s and s to s So, the time s to 1s is marked as having an event