Effect of Phasor Measurement Unit (PMU) on the Network Estimated Variables

Full Paper
ACEEE Int. J. on Electrical and Power Engineering, Vol. 4, No. 1, Feb 2013

Effect of Phasor Measurement Unit (PMU) on the
Network Estimated Variables
Jitender Kumar*1, J.N.Rai2, Vipin3 and Ramveer S. Sengar4
1
Electrical Engineering Deptt, G L Bajaj Instt of Tech & Management, Gr. Noida, India
*corresponding author: jitender3k@gmail.com
2
Electrical Engineering Department, Delhi Technological University, Delhi, India
3
Mechanical Engineering Department, Delhi Technological University, Delhi, India
4
Electrical Engineering Deptt, Accurate Instt of Management & Tech., Gr. Noida, India
Email: jnrai.phd@gmail.com, vipin2867@yahoo.co.in, ramveerchiro@gmail.com

Abstract - The classical method of power measurement of a and classical technique. The natural approaches of parameter
system are iterative and bulky in nature. The new technique measurement will extravagance the PMU as computational
of measurement for bus voltage, bus current and power flow is burden on measurement. The optimal placement of PMU
a Phasor Measurement Unit. The classical technique and PMUs devices is a difficult task for power system state estimation is
are combined with full weighted least square state estimator
investigated, in the literature. This paper will reflect the effect
method of measurement will improves the accuracy of the
measurement. In this paper, the method of combining Full
of PMU on the accuracy of measurement on state estimation
weighted least square state estimation method and classical parameters. In case 1, the classical state estimation method
method incorporation with PMU for measurement of power without using any PMU. But in case P, the measurement of
will be investigated. Some cases are tested in view of accuracy state estimation with PMU only is discussed.
and reliability by introducing of PMUs and their effect on
variables like power flows are illustrated. The comparison of II. WLS STATE ESTIMATION METHOD
power obtained on each bus of IEEE 9 and IEEE 14 bus system
will be discussed. Consider the set of measurements given by the vector z:-
z  h( x)  e (1)
Keywords – classical method; phasor measurement units
(PMUs); state estimation; full weighted least square (WLS) Where:
state. h T  [ h 1 ( x ), h 2 ( x ), h 3 ( x ),..... h m ( x )] (2)
I. INTRODUCTION hi (x) is the non-linear function relating measurement i to the
state vector x
The nature of electrical waves on grid will reflect the xT = [ x1, x2, x3............ xn] is the state vector of system
health of the system. The electrical wave is consisting of a eT = [e1 e2 e3……..em ] is the error measurement in state vector.
complex number that represents both the magnitude and The full weighted least square estimator [1] [3] [6] will
phase angle of the sine waves. The “Synchrophasor” is a minimize the following objective function:
device which measure nature of waves with respect to time
[5] [10]. The PMU is a device that allows the measurement of
m
(zi  hi (x))2
J (x)    [z  h(x)]T R 1[z  h( x)] (3)
voltage with multiple current options at each bus. The recently i 1 Rii
developed PMU [7] [8] [9] will help in deciding to stall power
improvement devices at proper location for:- At minimum value of objective function, the first-order
 Comprehensive planning optimum condition to be satisfied. It can be expressed as
 More accurate follows:
 Congestion tracking, J ( x )
 Advanced warning systems, g ( x)    H T ( x ) R 1 [ z  h ( x )]  0 (4)
x
 Information sharing,
The Taylor series of non-linear function g(x) can be
 Enhancement of System Integrity Protection
expanded for the state vector xk by neglecting the higher
Schemes (SIPS),
order terms [2] will be as
 Quick restoration of Grids,
 Effective grid operation, g ( x)  g ( x k )  G ( x k )( x  x k )  .......  0 (5)
 Dynamically manage the grid. The Gauss-Newton method is used to solve the above
The classical method of measurement of power flow and equation:
voltage on a bus of power system are iterative and bulky in x k+1= xk – [G(xk)] -1 . g(xk) (6)
nature. The full weighted least square state technique is a where, k is the iteration index, xk is the state vector at iteration
nonlinear in nature, but with first order Taylor series it become k and G(x) is called the gain matrix and expressed as:
a linear. Formerly conducted research will help in formulation
of a relation between full weighted least square state technique
© 2013 ACEEE 46
DOI: 01.IJEPE.4.1. 2

Full Paper

g ( x k ) IV. CLASSICAL METHOD WITH WLS
G ( x)   H T ( x k ) R 1 H ( x k ) (7)
x A PMU device will measure one voltage with multiple
k T k 1 k
(8) current phasors on a bus. Figure 1 shows a 4-bus system
g (x )  H (x ) R [ z  h ( x )]
example which has single PMU at bus 1. It consist of one
Normally, the gain matrix is sparse matrix and decomposed voltage with three current phasor measurements, namely as
into triangular factors. At each iteration k, the gain matrix are
V1  θ1, I1  δ1, I2  δ2 and I3  δ3
solved by using forward / backward substitutions, where
x k 1  x k 1  x k and
[G( xk )] xk 1  H T ( xk )R1 [ z  h( xk )]  H T ( xk ). R1 zk (9)
These iterations are going on till the maximum variable
k
difference satisfies the condition, ‘ Max x   ‘.

III. CLASSICAL METHOD
The traditional/classical method of measurement is Figure 1. Single PMU Measurement Model
consider current as a relation of power flow with respect to
The transmission line normally defined as pie network
bus voltages as
due to their advantages on system constraints. If y is defined
2 2 2 2 as the series admittance and yshunt as shunt admittance then
I ij  ( g ij  bij ) (V i  V j  2Vi V j Cos  ij ) the current measurements can be in rectangular coordinates
2 2 as in Fig 2.
Pij  Qij (10)

Vi
The power injection at bus i can be expressed as,
S i  Pi  jQi (11)
N
Pi  Vi  j 1
V j (Gij cos  ij  Bij sin ij ) (12)

N
Qi  Vi j 1
V j (Gij sin ij  Bij cos ij ) (13) Figure 2. Transmission Line in ‘π’ Model
The expressions for current in real and imaginary
The power flow from bus i to bus j are, component are as:-
S ij  Pij  jQij (14) Cij  Vi Ysi cos(i si )  Vj Yij cos( j ij )  Vi Yij cos(i ij ) (18)
2
P  Vi (gsi  gij )  Vi Vj (gij cosij  bij sinij )
ij (15) Dij  Vi Ysi sin( i si )  Vj Yij sin( j ij )  Vi Yij sin( i ij ) (19)
  
2 where, state vector is specified as:
Qij   Vi (bsi  bij )  Vi Vj (gij sinij  bij cosij ) (16)
x  [ V1 0 0 , V2  2 , V3  3 ..........VN  N ] T (20)
So the Jacobian H matrix will be as
The Jacobian H matrix corresponding to their real and
reactive parts is:
  P inj  P inj 
   Cij
 P V   Ysi cos( i   si )  Yij cos( i  ij ) (21)
 flow  P flow

Vi
  V 
  Q inj  Q inj  C ij
   Yij cos(  j   ij ) (22)
H    V  V j
  Q flow  Q flow 
  V  (17) Cij
  I mag  I mag    Vi Ysi sin(i  si )  Vi Yij sin(i  ij ) (23)
  i
  V 
  V mag   C ij
0   V j Yij sin(  j   ij )

 V 
 (24)
47
 j
© 2013 ACEEE
DOI: 01.IJEPE.4.1.2

Full Paper

Dij ˆ
E T T  A
 Ysi sin( i   si )  Yij sin( i   ij ) (25) x     ( H new R 1 H new ) 1 H new R 1  
ˆ (36)
Vi ˆ B
F 
 Dij ˆ
So that equation for x can be written in rectangular forms
 Yij sin(  j   ij ) (26) of z vector and H matrix. They are all in real numbers. In
V j
respect of the system, the PMU can deliver more precise
Dij information about system parameters. Some cases to be
 Vi Ysi cos(i  si )  Vi Yij cos(i  ij ) (27) performed on classical measurement set with and without
i PMUs. The different cases simulations and analysis are as
shown in Table I with some IEEE bus systems in the next
 D ij section.
 V j Yij cos(  j   ij ) (28)
 j TABLE I. D IFFERENT CASES IN IEEE SYSTEM
The measurement vector z will be
T T T T T T T
z  [ Pinj , Qinj , Pflow , Q flow , V ,  T , Cij , Dij ]T (29)
Generally, measurements attain by PMUs are more
accurate as compared to the traditional measurements. So
that measurements performed with PMUs are projected to be VI. SIMULATION RESULTS
more precise and accurate as estimated by classical methods.
Some cases are tested for analyzing the system variables
V. STATE ESTIMATION WITH PMUS accuracy with or without PMU, with the help of MATLAB
simulink software. The PMU locations in IEEE 9 Bus System
The state vector can be expressed as in rectangular and IEEE 14 Bus System at specific Bus Number are as shown
coordinates. The voltage measurement ( V  V  ) can be in Table II.
state as (V = E + jF), and the current measurement can be TABLE II. PMU LOCATIONS FOR EACH IEEE SYSTEM
state as (I = C + jD). Where series admittance of the line as
( gij + jbij ) and shunt admittance of the transmission line as
(gsi ??+ jbsi ). The flow of line current Iij can be expressed as:-
Iij  [(Vi  V j )  ( gij  jbij )]  [Vi  (gsi  jbsi )] The circuit diagram will be shown as in Figure 3 for IEEE
(30) 9 bus system and Figure 4 for IEEE 14 bus system.
 Vi  [(gij  jbij )  ( gsi  jbsi )]  V j  ( gij  jbij )
The vector z is state as z = h(x) + e, (where x is a system state
vector, h(x) is a linear equations, and e is an error vector). In
rectangular coordinates:
z = ( Hr + jHm )( E + jF ) + e (31)
where, H = Hr + jHm , x = E + jF and z = A + jB.
A and B are expressed by:
A = Hr  E – Hm  F (32)
B = Hm  E + Hr  F (33)
In matrix form,

 A  H r  H m  E 
 B    H H r  F 
e (34)
   m  

ˆ ˆ ˆ
Then, the estimated value x  E  jF can be solved as:-

x  ( H T R 1H ) 1 H T R 1z  G 1H T R 1z
ˆ (35)
If we define the linear matrix Hnew as

 H  Hm  Figure 3. IEEE 9 Bus System
H new   r
 H m H r  , then the eq. (35) can be written
 In this segment, IEEE bus systems as IEEE 9 bus system
[4] and IEEE 14 bus system [11] are tested with their respective
as: cases to find out the consequences of the PMUs to the
precision of the estimated variables.
© 2013 ACEEE 48
DOI: 01.IJEPE.4.1. 2

Full Paper

The graph to be designed on the basis of Standard
Deviation (S D) variation in each parameter’s on each Bus
which is planned for 20 cycle of operation of each IEEE
system, it has standard deviation categorized in two
categories i.e. Minimum & Maximum variation on actual
parameters received at each Bus. The figure 5 & figure 7 will
illustrate the variation in Standard Deviation at Minimum &
Maximum of Real Power (P) with respect to Each Bus on

Figure 6. Graph P (SD) vs Bus Number

Figure 4. IEEE 14 Bus System
Assume n as the number of variables, m as the number of
measurements and ε as the ratio of the number of
measurements per the number of variables. During the tests,
maintained ε as 1.6. Table III has more detailed information
about the measurement numbers for the tests [12].
TABLE III. VARIABLE T YPE AND MEASUREMENTS

A PMU has much smaller error deviations than classical
measurements as 0.0000001.The parameters measured are real
power (flow & injected) and reactive power (flow & injected)
measurements. The variation of parameters with or without
PMU easily reflected in the Figure 5 – 12 as given below:

Figure 5. Graph P (SD) vs Bus Number Figure 9. Graph Q (SD) vs Bus Number
© 2013 ACEEE 49
DOI: 01.IJEPE.4.1.2

Full Paper

The table IV illustrates that how the Standard Deviation
(S D) enhanced at each stage as PMU increase from No PMU
to Only PMUs. In IEEE 14 bus system, the S.D. of the
estimated Power is approximately 1.88E-05 when there is no
PMUs, but at Only PMUs it become 1.16E-01. It means that
the S.D. of ‘No PMUs’ is increased by adding PMUs in system.
The interesting thing is that the standard deviation increasing
as increasing PMU. Therefore, this result demonstrates that
the effectively installing of PMUs will reduce the chances of
error in measurement of estimated variables.
Figure 10. Graph Q (SD) vs Bus Number

CONCLUSION
Now, the classical measurement method with the PMU is
able to measure the voltage and current with their magnitude
and phasors. The current measurement is implemented but
the measurement set as in the rectangular form. The Jacobian
matrixes are illustrated in detail for the measurements which
include the elements of Equations. The state estimation in
the linear formulation is investigated with PMU. All the
variables and their respective measurements are improved as
all in rectangular form, and then treated separately during the
Figure 11. Graph Q (SD) vs Bus Number estimation process. Such linear formulation of the PMU data
can produce the estimation result by a single calculation
without performing the any bulky iteration as in classical
methods. If only PMU data measurement set is exist in the
real measurement world, and then there will be improvement
in the computation time and accuracy estimation as compare
to the linear formulation of the state estimation.
The advantages of using PMU will advances the accuracy
of the estimated variables. Some cases are tested while
gradually increasing the number of PMUs which are added
to the measurement set. With the help of advanced accuracy
of PMU, it was seen that the estimated accuracy is also
Figure 12. Graph Q (SD) vs Bus Number
increases. One of the motivating thing is that the accuracy
varying the number of PMU accretion from ‘No PMU (Case improves most effectively when the number of implemented
1)’ to ‘Only PMU (Case P)’ for IEEE 9 bus system. Similarly PMUs are around ‘14 %’ of the system buses. It is proved
Figure 6 & Figure 8 will illustrate the variation in Standard that the quality of the estimation is enhanced by adopting
Deviation at Minimum & Maximum of Real Power (P) with PMU data to the set of measurement. The PMU
respect to Each Bus on varying the number of PMU accretion measurements will provide us improved accuracy and
from ‘No PMU (Case 1)’ to ‘Only PMU (Case P)’ for IEEE 14 redundancy for the system.
bus system. The Standard Deviation in Reactive Power (Q) The study is carried out to establish a relationship among
will be illustrated in Figure 9 – 12 from ‘No PMU (Case 1)’ to classical method, full weighted least square state estimation
‘Only PMU (Case P)’ for IEEE 9 bus system and IEEE 14 bus method and Phasor Measurement Units. Further it is visualized
system. that the information of power variation on each bus of system
TABLE IV. AVERAGE S.D. IN ESTIMATED VARIABLES are more precise and accurate as compared to the classical
method. The linear formulation will suggest us a more specific
and accurate information of power variation without doing
any bulky iteration as performed in the classical methods.
The use of PMUs will lead us to enhanced accuracy in results.

REFERENCES
[1] Abur and Exposito A. G., Power System State Estimation,
Theory and Implementation, MAECEL DEKKER, 2005, pp.
9-27.
[2] Chakrabarti S. and Kyriakides E., “Optimal Placement of
Phasor Measurement Units for Power System

© 2013 ACEEE 50
DOI: 01.IJEPE.4.1.2

Full Paper

Observability”, IEEE TRANSACTIONS ON POWER [7] Phadke A. G., “SYNCHRONIZED PHASOR
SYSTEMS, VOL. 23, NO. 3, AUGUST 2008, 0885-8950 © MEASUREMENTS – A HISTORICAL OVERVIEW”, 0-
2008 IEEE, pp.1433–1440 7803-7525-4/02 © 2002 IEEE, pp.476–479
[3] Cheng Y., Hu X., and Gou B., “A New State Estimation using [8] Phadke A.G, Thorp J.S., Nuqui R.F. and Zhou M., “Recent
Synchronized Phasor Measurements”, 978-1-4244-1684-4/08 Developments in State Estimation with Phasor
©2008 IEEE, pp.2817–2820 Measurements”, 978-1-4244-3811-2/09 ©2009 IEEE, pp.1-7
[4] Ebrahimpour R., Abharian E. K., Moussavi S. Z. and Birjandi [9] Rahman K. A., Mili L., Phadke A., Ree J. D. L. & Liu Y.,
A. A. M., “Transient Stability Assessment of a Power System “Internet Based Wide Area Information Sharing and Its Roles
by Mixture of Experts”, INTERNATIONAL JOURNAL OF in Power System State Estimation”, 0-7803-6672-7/01 © 2001
ENGINEERING, (IJE) Vol. 4, March 2011 pp.93–104. IEEE, pp.470–475
[5] Klump R., Wilson R.E. and Martin K.E., “Visualizing Real- [10] Real time dynamics monitoring system [Online]. Available:
Time Security Threats Using Hybrid SCADA / PMU http://www.phasor rtdms.com
Measurement Displays”, Proceedings of the 38th Hawaii [11] Tanti D.K., Singh B., Verma M.K., Mehrotra O. N., “An
International Conference on System Sciences – 2005, 0-7695- ANN based approach for optimal placement of DSTATCOM
2268-8/05 ©2005 IEEE, pp.1–9 for voltage sag mitigation”, INTERNATIONAL JOURNAL
[6] Madtharad C., Premrudeepreechacham S., Watson N. R. and OF ENGINEERING SCIENCE AND TECHNOLOGY
Saenrak D., “Measurement Placement Method for Power (IJEST), ISSN : 0975-5462 Vol. 3 No. 2 Feb 2011 pp.827–
System State Estimation: Part I”, 0-7803-7989-6/03 ©2003 835
IEEE, pp.1629–1632 [12] Xu B. and Abur A., “Observability Analysis and Measurement
Placement for Systems with PMUs”, 0-7803-8718-X/04 ©
2004 IEEE, pp.1–4

© 2013 ACEEE 51
DOI: 01.IJEPE.4.1.2

Effect of Phasor Measurement Unit (PMU) on the Network Estimated Variables

More Related Content

What's hot (16)

Viewers also liked (20)

Similar to Effect of Phasor Measurement Unit (PMU) on the Network Estimated Variables (20)

More from IDES Editor (20)

Recently uploaded (20)

Effect of Phasor Measurement Unit (PMU) on the Network Estimated Variables