Predictive Out-of-Step Protection and Control Scheme

1

• Predictive Out-of-Step Protection and Control
  Scheme
• Based on Real-Time
• Phasor Measurement Application
• By Alla Deronja, P.E.
• Senior System Protection Engineer
• American Transmission Company
• Milwaukee, Wisconsin

2

• This presentation is focused on demonstration of
  applying
• the concept of real-time measurements in a
  transmission
• system to perform system protection and control
  functions
• in a predictive - or adaptive - manner to account
  for the
• system operating conditions at any time.
• Objectives
• Real-time measurements and adaptive/predictive
  protection.
• Predictive out-of-step protection scheme based
  on real-time measurements.
• Practical application.

3

• As a system protection engineer, I am interested
  in
• applying real-time phasor measurement technology
  to
• perform system protection, control and tripping
  functions
• to achieve a protection systems adaptability to
  account
• for all possible system operating conditions.

4

• The concept of adaptive relaying is that many
  relay
• settings are dependent upon assumed conditions in
  the
• power system. In order to cover all possible
  scenarios that
• the protection system may have to face, the
  actual
• protection settings are often not optimal for any
  particular
• system state. If an optimal setting is desired
  for an
• existing condition on the power system network,
  then it
• becomes necessary for the setting to adapt itself
  to the
• real-time system states as the system conditions
  vary
• due to changing loads, network switching
  operations,
• or faults.

5

• A number of relaying functions are predicated
  upon an
• assumed pattern of power system behavior during
• transient stability oscillations and other
  dynamic
• conditions. An example of such functions is
  out-of-step
• protection.
• A predictive out-of-step protection and control
  scheme
• based on the real-time phasor measurement
  principle that
• has inspired me to pursue my project proposal has
  
• been already in operation for almost 20 years.

6

• TOKYO ELECTRICS PREDICTIVE OUT-OF-STEP
  PROTECTION SYSTEM
• The predictive out-of-step protection system by
  means
• of observing phase differences between power
  centers
• and based on the realtime phasor measurement
• principle similar to the one I sought to install
  at my
• company has been installed and successfully
  operated
• by Tokyo Electric Power Company (TEPCO, Japan)
  since
• February, 1989.

7

• The bulk power system of TEPCO in presented in
  Fig.
• 1. A major characteristic of this power system is
  that
• the power generation areas are far from the
• consumption areas. The eastern, northern, and
• southeastern generator groups are linked by a
  bulk
• power system comprising 500 kV double-circuit
• transmission lines configured in duplex and
  triplex
• routes and forming a trunk. The western
  generator
• group and large local loads are thus linked to
  the bulk
• power system via the substations marked as A1 and
  A2
• in Fig. 1.

8

• FIGURE 1. BULK POWER SYSTEM OF TEPCO

9

• The western generator group tends to be heavily
• loaded, and its own capacity cannot meet demand.
• Power is received from the bulk power system to
  make
• up the deficiency.
• When a double fault occurs along both circuits of
  a
• double-circuit line forming one route, the
  substations at
• both ends of the line are disconnected and
  transmission
• capability is interrupted. If a successive fault
  occurs
• after reclosing, a slow cyclic power swing
  develops
• between the western generator group and the bulk
• power system.

10

• The same situation occurs in the event of failure
  of a
• bus-bar protective relay to operate during a
  bus-bar
• fault. Over time, the phase difference of the
  generator
• groups thus undergoes oscillating divergence. If
  this
• condition is not corrected, an out-of-step
  condition will
• begin to occur in various parts of the power
  system and
• may lead to its total collapse.

11

• In order to maintain the reliability of the power
  system
• should such serious rare faults occur, a
  predictive
• protection system that prevents total collapse of
  the
• power system has been developed. This protection
• system utilizes online data collected before and
  after
• the onset of a system disturbance to determine
• the characteristics of the power swing and
  predict an
• out-of-step condition. The system operates during
  the
• incubation period so that appropriate control can
  be
• performed before the out-of-step condition occurs.

12

• The western area can be islanded (Fig. 2) from
  the bulk
• power system at points (a) or (b) and (a) or
  (b) before
• out-of-step occurs and then operated
  independently.
• This eliminates power swing between the generator
  
• groups of the two systems and restores stability.
  
• The separation point is selected based on the
  power
• flow at pre-determined points for separation
  before the
• fault.

13

• FIGURE 2. SIMPLIFIED MODEL OF TRUNK LINE
• AND GENERATORS

14

• The protection scheme is outlined in Fig. 3. The
  status
• of each generator group (east, north, southeast,
  west)
• is obtained by measuring the bus-bar voltages of
• neighboring substations as representative values.
  From
• these, the phase differences between the western
• generator group and the bulk power system are
• obtained. From the phase difference values, the
• corresponding values for 200 ms in the future are
  
• predicted. If the latter exceed the setting value
  (?limit),
• the respective generator groups are judged to be
• unstable.

15

• FIGURE 3. OUTLINE OF PROTECTION SCHEME

16

• The two out of three logic is employed to judge
• instability of all the groups and to prevent
  unwanted
• operation in the event of an out-of-step of one
• generator group within the bulk power system. In
  order
• to initiate system separation, the current swing
• detection element must operate based on the
  current
• flowing through the transformers of the linkage
• substations A1 and A2 between the bulk power
  system
• and the western area (Fig. 1).

17

• When the two out of three generator groups have
  been
• determined to be unstable and the current swing
• detection element has operated, the islanding
  command
• is issued from the Central Equipment to the
  separation
• point. The trip command is executed if the
  current
• swing detection element in the RTU has operated
  on
• the current flowing through that point. The
  current
• swing detection element is provided as a
  fail-safe
• measure for the protection calculation based on
  phase
• difference.

18

• To achieve flexibility in dealing with various
  operating
• conditions of the western area, the present
  protection
• system is divided into completely separate duplex
  
• systems System 1 and System 2. System 1 handles
• separation at points (a) and (b) in Fig. 2, and
  System 2
• at points (a) and (b).
• System 2 is nearly identical to System 1 so that
• System 1 will be only described.

19

• System 1 comprises a Central Equipment (Phasor
  Data
• Concentrator) at Substation A1 and Remote
  Terminal
• Units RTUs (Phasor Measurement Units PMUs) at
  
• Substations B1, C, D, and E (Fig. 4).
• The RTUs simultaneously sample bus-bar voltages
  at
• 600 Hz. The sampled real-time data is transmitted
  to
• the Central Equipment (CE) via the data
  transmission
• system (a communication channel). In addition,
  the
• current flowing through the system separation
  point (b)
• at Substation B1 is measured then, the power
  flow
• value is calculated and transmitted to the CE.

20

• FIGURE 4. BASIC CONFIGURATION OF SYSTEM 1
• PROTECTION SYSTEM

21

• At substation A1, the current and power flow
  values
• through the system separation point (a) and
  through
• the transformer are measured. The CE calculates
  the
• real-time phase difference and predicted future
  phase
• difference and selects the system separation
  point using
• this online data. The circuit breaker command is
  issued
• for either point (a) or (b) as selected by the CE
  
• calculation on the measured data.

22

• The entire system configuration is presented in
  Fig. 5.
• All devices are based on 16-bit microprocessors.
  The
• microprocessor has data transmission,
  self-diagnostic
• and calculating functions provided by the CE.
  Data
• transmission between the CE and RTUs is carried
  out
• synchronously via a duplex digital microwave
  link. The
• data transmission speed of this system is 56 kbps.

23

• FIGURE 5. OVERALL CONFIGURATION
• OF
  PROTECTION SYSTEM

24

• The data from one RTU of the C, D, and E
  substations
• is sent to both CE of the same system via the
  data
• transmission system. Each CE receives the data
  for both
• RTUs of each generator group, but only one set of
  data
• is normally employed. If an abnormality occurs,
  the CE
• can switch to the other RTUs data. The purpose
  of this
• two-fold redundancy is to decrease system
  downtime
• since long-distance transmission with multiple
  spans is
• employed.

25

• The method of obtaining the phase difference
  between
• two points (western generator group and bulk
  power
• system) from simultaneously sampled voltage data
  is as
• follows.
• In Fig. 6, representative voltage waveform
  sampling
• values for a bus-bar in the vicinity of the
  northern
• generator group (C in Fig. 6) and the western
• generator group (B in Fig. 6) are shown.

26

• FIGURE 6. EXAMPLE OF VOLTAGE WAVEFORM SAMPLING
• DATA FOR TWO SUBSTATIONS

27

• The phase difference, ?n, at present time n can
  be
• obtained from the voltage data VBn, VBn-3, VCn,
  and
• VCn-3 for the present time and three previous
• samples as follows

28

• Thus,
• In order to simplify the calculation by replacing
  X Xi
• with a first-order approximation obtained via
  Taylor
• Series, the phase difference can be obtained by
• where 0 ? X ? 1

29

• If X ? 1, the following additional equation is
  used to
• perform the calculation.
• If X ? 0, the following additional equation is
  used to
• perform the calculation.
• Since the approximation error is on the order of
  10-2,
• accuracy is sufficient for practical use.

30

• The phase difference ? between the two areas can
  be
• approximated by the following equation.
• where,
• ?0 is initial value of phase difference ?,
• ? is angular frequency of ?,
• ? is damping constant,
• A is amplitude.
• This equation interprets the power swing mode as
  a
• sine wave that diverges or converges.

31

• Using the phase difference values for the present
  time
• and previous time, the future phase difference
  value
• can be predicted. The predicted phase difference
  ? for
• time TH in the future is derived from eight
  pieces of
• data (Fig. 7) and calculated as follows

32

• FIGURE 7. METHOD OF PREDICTING
• PHASE
  DIFFERENCE

33

• Values of 200 ms and 100 ms were selected for TH
  and
• TK (a time interval before the present time in
  Fig. 7),
• respectively, in order to predict accurately and
  to
• provide an acceptable operating time.
• A simulation of the present prediction algorithm
• calculation is given in Fig. 8. The results agree
  well
• with the present phase difference value and the
• predicted future value at 200 ms.

34

• FIGURE 8. SAMPLE CALCULATION OF
• PHASE DIFFERENCE

35

• When the obtained predicted phase difference
  value ?
• exceeds the setting value ?limit, it is judged
  that the
• power swing between the two generator groups is
• unstable. The value for ?limit is determined by
  computer
• simulation under varying conditions and must
  guarantee
• operation when the system is unstable and prevent
  
• operation when the system is stable.
• The table in Fig. 9 shows the results of computer
  
• simulation for several system operating patterns.
  ?limit is
• set to 100?.

36

• FIGURE 9. SIMULATION RESULTS

37

• In order to guarantee fail-safe operation of the
  scheme,
• an input different from the voltage input, the
  current
• input, is used to detect the swing. Its logic
  shown in
• Fig. 10 comprises a rate of change detection
  block to
• determine whether power swing is present and a
• magnitude of change detection block to detect the
  size
• of the current fluctuation. The element operates
  on the
• AND of these two blocks.

38

• FIGURE 10. CURRENT SWING DETECTOR

39

• The operation of the current swing detector is
• presented in Fig. 11. The magnitude of change
• detection block operates when the measured
  current
• fluctuation magnitude is greater than its
  sensitivity
• setting ISET during maximum power swing period
  ?Tmax.
• The values for ISET and ?Tmax are determined by
• computer simulation, and ISET is set for ?Tmax3
  sec.

40

• FIGURE 11. PRINCIPLE OF CURRENT SWING DETECTION

41

• The element detects when the slope of the
  difference of
• the r.m.s. current value during the small
  interval ?t
• (?I/?t) is greater than constant K and continues
  longer
• than operation delay time T1 in order to operate
• when the size of the current swing is greater
  than ISET
• during a sine wave that is smaller than ?Tmax. To
  
• prevent dropout in the vicinity of extreme values
  of the
• sine wave, the OFF delay timer T2 (reset delay)
  is set to
• 1 sec when ?t40 ms and T1200 ms.

42

• A computer simulation was run for a double three-
• phase-to-ground fault of a double circuit
  transmission
• line (1 route) in the trunk with a subsequent
• failure of three-pole reclosing with synchronism
  check.
• The results are presented in Fig. 12.
• The power swing has a tendency toward divergence
• without a protection system (Fig. 12a). In this
  case,
• out-of-step is likely after 10 sec, and this
  process would
• continue to extend and eventually cause a
  conventional
• out-of-step relay to operate and separate the
  western
• area after about 13.5 sec.

43

• FIGURE 12. EVALUATION BY COMPUTER SIMULATION

44

• Even after separation, the power swing would also
• continue in the bulk power system.
• In contrast, when the protection system is
  employed
• as shown in Fig. 12b, the western area is
  separated
• after 6.6 sec, and the power swing of the bulk
  power
• system starts to converge. The western area is
  then
• operated independently, and another control
  system
• such as under-frequency load shedding adjusts the
  
• power supply and demand balance.

45

• The present protection system also underwent a
  field
• test. The power system configuration and test
  results
• are presented in Fig. 13. A circuit breaker at
  point (c) is
• being closed to configure the western power
  system as
• a loop system.The predicted and measured phase
• difference values were observed. The predicted
  phase
• difference values before and after closing of the
  circuit
• breaker agreed well with the measured values.

46
b) Example of phase difference swing detection

• FIGURE 13. CONTENT AND RESULTS OF FIELD TEST

47

• EXAMPLE OF
• OUT-OF-STEP PROTECTION SCHEME
• POSSIBLE UPGRADE UTILIZING
• SYNCHROPHASOR MEASUREMENTS
• The company I am representing - American
  Transmission
• Company - utilizes an out-of-step protection
  scheme,
• which I sought to upgrade to make it adaptive -
  or
• predictive - using the real-time synchrophasor
• measurement principle and newest hardware
  available.

48

• The existing ATC Northern transmission system has
  
• limited power transfer capabilities from
  Michigans Upper
• Peninsula (UP) generation to a key bulk power
• transmission substation due to its inadequate
  transmission
• and generation infrastructure.
• Several completed major projects have improved
  the
• system. However, it remains weak and still
  requires the
• use of a special protection scheme to block
• unstable power swings by tripping UP generation
  for
• critical faults on the transmission system.

49

• Two pairs of power swing relays SEL-68 are
  installed at
• the key bulk power transmission substation to
  backup the
• UP generations SPS operations. Each pair is
  installed to
• trip in series for redundancy to prevent a
  misoperation
• should one of the relays fail.
• The power swing relays are time-delayed to allow
• operation of the primary SPS. If the primary SPS
  fails to
• operate, the power swing relays are designed to
  separate
• the ATC power system into two islands the UP and
  the
• Wisconsin bulk power system.

50

• For unstable power swings to the south of the key
  bulk
• power transmission substation caused by an excess
  of the
• UP generation, one pair of the SEL-68 relays will
  trip the
• three southern lines of the power transmission
  corridor.
• For unstable power swings to the north of the
• substation caused by a deficiency of the UP
  generation,
• the second pair of the SEL-68 relays will trip
  the three
• northern lines of the power transmission corridor.

51

• FIGURE 14. CURRENT ATC POWER SWING RELAY SCHEME

52

• The existing out-of-step protective relays are
  SEL-68,
• the 1987 years vintage. Their principle of
  operation is
• based on comparing in the central location the
  voltage
• phasor angles on two buses that present two
  different
• systems which, if not synchronized, will have to
  be split
• by the out-of-step relaying.

53

• FIGURE 15. POWER SYSTEM
  MODEL

54

• DELTA (?) is the angle between source E1 and
  source
• E2, and V and I are the relays voltage and
  current,
• respectively.
• E1VjX1I
• E2VjX2I
• The angle DELTA between these phasors is found as
  
• follows
• AjBE1(E2)

55

• Where A is the real part of the complex product
  of E1
• and the complex conjugate of E2
• B is the imaginary part.
• Then, DELTA inverse cotangent (A/B)
• DELTA-double-dot is found from DELTA-dot by
  filtered
• first derivatives.

56

• DELTA, DELTA-dot, and DELTA-double-dot are used
  by
• the relay SEL-68 to assess system stability by
  examining
• their loci in two planes (figures 16 and 17).

57

• FIGURE 16. DELTA VS. DELTA-DOT PLANE

58

• FIGURE 17. DELTA-DOT VS. DELTA-DOUBLE-DOT
  PLANE

59

• If tripping (or blocking) is initiated when the
  swing
• trajectory crosses the TRIP-BLOCK DECISION LINE
  in
• the DELTA vs. DELTA-dot plane (Fig. 16), then the
  
• power circuit breaker has just the necessary time
  to
• complete its operation before the TRIP-BLOCK
• DECISION ANGLE (TBDA) is actually reached.
• In Fig. 17, the loci for several different stable
  () and
• unstable () swings are plotted at the moment
  that the
• power angle crosses the TRIP-BLOCK DECISION LINE.

60

• This explains how information from the two planes
  
• (figures 16 and 17) is utilized by the SEL-68
  relay to
• arrive at a TRIP-BLOCK (or, equivalently, STABLE-
• UNSTABLE) decision. Examining the swing loci in
  the
• DELTA vs. DELTA-dot plane permits the relay to
• determine when the power angle has reached a pre-
• determined value (TBDA), and plotting the loci in
  the
• DELTA-dot vs. DELTA-double-dot plane at that
  instant
• permits the stable/unstable decision.

61

• The relays are designed to support four system
• operating conditions, one normal and three
  alternative,
• each based on pre-calculated system impedances.
• It is clear that the power system can have more
  than
• four operating conditions, and pre-calculated
  impedance
• values, even updated on a regular basis, will
  never
• provide the most accurate representation of the
  real
• system configuration. A precedence for potential
• misoperation or insecurity of the protection
  system has
• been set.

62

• Upon detection of an out-of-step condition, the
  next
• step of the protection system is to permit
  selective
• tripping for clearly unstable cases so that the
  power
• system is separated in the islands with a
  reasonable
• match between load and generation in each island.
  At
• present, the places where tripping is permitted
  are pre-
• determined based upon simulations performed
  during
• system stability studies. Eventually, a more
  appropriate
• procedure may be developed to determine both the
• nature of an in-progress swing as well as
  desirable
• locations for separation in real-time.

63
FIGURE 18. PREDICTIVE OUT-OF-STEP
PROTECTION SYSTEM
BASED ON REAL-TIME PHASOR MEASUREMENTS
64

• Implementation of this scheme will require
  substantial
• communication channel capacity. Phasors from the
  two
• key places must be communicated to a central
  location
• where stability swing evaluation and prediction
  are to
• be carried out. After the prediction phase,
  direct trip
• or block commands must be communicated locally
  and
• to one remote substation. It seems almost certain
  that
• fiber optic communication channels will be
  necessary
• for adaptive features of this category to be
• implemented.

65

• REFERENCES
• Y. Ohura, M. Suzuki, K. Yanagihashi, M. Yamaura,
  K. Omata, T. Nakamura, A predictive Out-of-Step
  Protection System Based on Observation of the
  Phase Difference Between Substations, IEEE
  Transactions on Power Delivery, Vol. 5, No. 4,
  November 1990.
• 2. A. G. Phadke and J. S. Thorp, Computer
  Relaying for Power
• Systems, Research Studies Press Ltd.
  ISBN 0 86380 074 2.
• 3. G. Benmouyal, E. O. Schweitzer III, A.
  Gusman, Synchronized Phasor Measurement in
  Protective Relays for Protection, Control, and
  Analysis of Electric Power Systems.

66

• REFERENCES (cont.)
• 4. Ph. Denys, C. Counan, L. Hossenlopp, C.
  Holweck, Measurement of Voltage Phase for the
  French Future Defense Plan Against Losses of
  Synchronism, IEEE Transactions on Power
  Delivery, Vol. 7, No. 1, January 1992.
• V. Centeno, J. De La Ree, A. G. Phadke, G.
  Mitchell, J. Murphy, R. Burnett, Adaptive
  Out-of-Step Relaying Using Phasor Measurement
  Techniques, IEEE Computer Applications in Power,
  October 1993.
• 6. S. H. Horowitz and A. G. Phadke, Boosting
  Immunity to Blackouts, IEEE Power Energy
  magazine, September/October 2003.

67

• REFERENCES (cont.)
• Schweitzer Engineering Laboratories, SEL-68 Out
  of Step Blocking/Tripping Relay Swing Recorder.
  Instructional Manual, April 1986
• 8. E. O. Scweitzer III, T. T. Newton, R. A.
  Baker, Power Swing Relay Also Records
  Disturbances, 13th Annual Western Protective
  Relay Conference, Spokane, WA, October 21-23, 1986