09/11/2022

Course Structure
Power system Planning and operation

EE 467
Planning Operational planning Operation
Power System Planning & Operation
• Long-term load • Automatic generation
• Short-term load
State Estimation forecasting
forecasting
control
• Generation expansion • Optimal power flow
• Unit commitment
• Transmission expansion • System security
Prof. Mohamed Abido • Economic dispatch
• Maintenance scheduling • State estimation
1
2

1 2

Power System Planning & Operation State Estimation

• What are the states?
• Do we need to measure all the
states?
State Estimation (SE) • What is the state estimation?
• What is the condition for state
estimation?
3 4

3 4

Objectives State Estimation: Basics

• Basic Definitions Idea:
Use measurements to figure out the “state” (that is,
• Power System Observability Voltage magnitudes and phase angles) of an
• State Estimation Objective operating power system
• State Estimation Challenges Measurements:
• State Estimation Methods • voltage magnitudes
• power,
• State Estimation of AC Networks • VAR,
• ampere-flow
5 6

5 6

1
 09/11/2022

State Estimation Network Observability

• Observability A power system What is the minimum
is observable if the number of measurements
measurements made on it for system observability?
allow determination of bus
voltage magnitude and angle
Network Observability at every bus of the network.

• At which bus the measuring

devices should be installed
for the system to be
observable?
7 8

7 8

Network Observability Network Observability

• What is the Given one measurement on a 3-bus
situation for system, would be the system observable?
this system? Can we estimate the two unknown
phase angles?
This network is “unobservable”.
• Optimization
Solution, add a pseudo measurement to make it
observable, for example use a load prediction program
to get an estimate of the load on bus 2 and add to the
measurement set. It is now observable.
9 10

9 10

State Estimation State Estimation

Problems/Challenges:
Solution:
• Measured quantities have noise and errors
A state estimator that can:
• Limited number of measurements, not as many as
necessary • smooth out small random errors in meter readings
• Measurement transducer errors • detect and identify gross measurement errors
• Communication failure
• fill in meter readings that have failed due to
• Some measurements are BAD (measured value is very communication failures.
different from the actual quantity)
11 12

11 12

2
 09/11/2022

State Estimation State Estimation

Approximations Approximations
• The power flow between bus k and bus j is • Reactive power flow across circuits is determined by the
𝑉𝑘 𝑉𝑗 difference in the voltage phasor magnitudes between the
𝑃𝑘𝑗 = sin 𝛿𝑘 − 𝛿𝑗 terminating buses. So, reactive power flow can be neglected.
𝑋𝑘𝑗
• In the per-unit system, the numerical values of voltage magnitudes • Then, the power flow can be approximated to
1
|Vk| and |Vj| are very close to 1.0. Typical range under most 𝑃𝑘𝑗 = 𝛿 − 𝛿𝑗
operating conditions is 0.95 to 1.05. So, |Vk|=|Vj|1.0 pu 𝑋𝑘𝑗 𝑘
• In addition, (θk – θj) is less than 10-15 degrees. So, sin 𝛿𝑘 − 𝛿𝑗 • That means, the real power flow across circuits is determined by
≅ 𝛿𝑘 − 𝛿𝑗 the difference in voltage phasor angles between the terminating
buses.
13 14

13 14

Example: Example:
Suppose we use accurate
How many meter readings values of two-meter
are required to calculate readings M13 and M32 and
the bus phase angles and M13 = 5 MW = 0.05 pu
all load and generation M 32 = 40 MW = 0.40 pu
values fully (observable)? Find all the states, loads,
generations, and line
The answer is 2 flows.
(Can U exercise it!)
15 16

15 16

Example: Example: Measurements with errors

M13 = 5 MW = 0.05 pu ? Suppose the readings obtained are:
M 32 =40 MW = 0.40 pu M12 = 62 MW = 0.62 pu
1 M13 = 6 MW = 0.06 pu
𝑓13 = 𝜃 − 𝜃3 = 𝑀13 = 0.05 𝑝𝑢
𝑋13 1 M32 = 37 MW = 0.37 pu
1
𝑓32 = 𝜃 − 𝜃2 = 𝑀32 = 0.40 𝑝𝑢 Using M13 and M32 , the phase
𝑋32 3
angles can be calculated as follows
Assume 3 = 0
1 = 002 rad Assume 3 = 0
2 = −010 rad 1 = 0024 rad
2 = −00925 rad
17 18

17 18

3
 09/11/2022

Example: Measurements with errors Example: Alternate measurements

1 = 0024 rad Use measurements at
2 = −00925 rad M12 and M32 result,
3 = 0 rad (still assumed zero) flows at M12 and M32
Using M13 and M32 the flows at match measured
those measurements match the values, but flow in line
measured values, but the flow 1 – 3 (6 MW) does not
on line 1 – 2 (62 MW) is not match M13 measured
equal to the M12 measured (7.875 MW)
value (58.25 MW).
19 20

19 20

Problem Objective Problem Statement

We need a procedure that uses the Redundant measurements can be used to
estimate the “states” (voltage magnitudes and
information available from all meters to phase angles) of the power system assuming:
produce the best estimate of the actual
• network configuration is known
angles, voltages, line flows, loads, and
• network impedances are known
generations.
• transformer taps positions are known
• phase angle regulator positions are known
21 22

21 22

State Estimation: Definition State Estimation: Techniques

State Estimation: is the art of estimating the The weighted least-squares criterion,
exact system state given a set of imperfect where the objective is to minimize the sum of
measurements made on the power system. the squares of the weighted deviations of the
estimated measurements, zˆ, from the actual
This leads to the problem of how to formulate a
measurements, z.
“best” estimate of the unknown parameters 𝑵𝒎
given the available measurements. 𝟐
𝑴𝒊𝒏𝒊𝒎𝒊𝒛𝒆 ෍ 𝒘𝒊 𝒛𝒊 − 𝒛ො 𝒊
Solution: Optimization 𝒊=𝟏
23 24

23 24

4
 09/11/2022

Weighted Least-Squares Estimator Matrix formulation of weighted least squares estimator

Estimating Ns unknown parameters (states: x1, x2, If the fi functions are linear, that is:
….,xns) using Nm measurements, we would write: 𝑓𝑖 𝑥1 , 𝑥2 , … , 𝑥𝑁𝑠 = 𝑓𝑖 𝒙 = ℎ𝑖1 𝑥1 + ℎ𝑖2 𝑥2 + ⋯ + ℎ𝑖𝑁𝑠 𝑥𝑁𝑠
𝑁𝑚 2 We can write
𝑍𝑖𝑚𝑒𝑎𝑠 − 𝑓𝑖 (𝑥1, 𝑥2 , … , 𝑥𝑁𝑠 )
min 𝐽 𝑥1 , 𝑥2 , … , 𝑥𝑁𝑠 = ෍
𝜎𝑖2 𝑓1 (𝒙)
𝑖=1
zi is the ith measured value 𝑓2 (𝒙)
𝒇 𝒙 = = 𝐻𝒙
fi is the estimated value (z˄) ⋮
σi is the weighting factor (meter’s accuracy is set as ±3σ, 𝑓𝑁𝑚 (𝒙)
for example, a meter with ±6 MW accuracy has σ = 2 H is NmNs matrix containing the coefficients of the linear
MW).
functions fi (x)
25 26

25 26

Matrix formulation of weighted least squares estimator Matrix formulation of weighted least squares estimator
In compact form, WLS can be written as: The solution of ∇ J(x) = 0 gives :
min 𝐽 𝒙 = 𝑍 𝑚𝑒𝑎𝑠 − 𝑓(𝒙) 𝑇 𝑅 −1 𝑍 𝑚𝑒𝑎𝑠 − 𝑓(𝒙)
(a) If Nm = Ns (Completely determined)
Where
𝑋 𝑒𝑠𝑡 = 𝐻 −1 𝑍 𝑚𝑒𝑎𝑠
𝑧1𝑚𝑒𝑎𝑠
𝝈𝟐𝟏 ⋯ 𝟎 (b) If Nm > Ns (Over-determined)
𝑧 𝑚𝑒𝑎𝑠
= 2
𝒛𝑚𝑒𝑎𝑠 𝑹 = ⋮ ⋱ ⋮ −1
⋮ 𝑋 𝑒𝑠𝑡 = 𝐻 𝑇 𝑅−1 𝐻 𝐻 𝑇 𝑅−1 𝑍 𝑚𝑒𝑎𝑠
𝟎 ⋯ 𝝈𝟐𝑵𝒎
𝑧𝑁𝑚𝑒𝑎𝑠
𝑚
R is a diagonal matrix. The minimum of J(x) is found when: (c) If Nm < Ns (Under-determined)
𝜕𝐽 𝒙
= 0, 𝑓𝑜𝑟 𝑖 = 1, … , 𝑁𝑠 OR ∇ J(x) = 0 𝑋 𝑒𝑠𝑡 = 𝐻 𝑇 𝐻 𝐻 𝑇 −1 𝑍 𝑚𝑒𝑎𝑠
𝜕𝑥𝑖
27 28

27 28

Example 1: WLS for 3-bus with 3 measurements system Example 1: WLS for 3-bus with 3 measurements system
Suppose the readings obtained are: Solution
M12 = 62 MW = 0.62 pu We will assume that the meter’s accuracy (± 3 MW) is being stated as equal
to the ± 3. Then ± 3 MW corresponds to a metering standard deviation of
M13 = 6 MW = 0.06 pu
 = 1 MW = 0.01 pu
M32 = 37 MW = 0.37 pu
Note that: Nm > Ns (over determined problem), then
Using M12 , M13 and M32 , find the best −1
estimate of the phase angles using WLS. 𝑋 𝑒𝑠𝑡 = 𝐻 𝑇 𝑅−1 𝐻 𝐻 𝑇 𝑅−1 𝑍 𝑚𝑒𝑎𝑠
Assume that all three meters have the 𝑋 𝑒𝑠𝑡= vector of estimated state variables
following characteristics. 𝐻 = measurement function coefficient matrix
Meter full-scale value: 100 MW Meter 𝑅 = measurement covariance matrix
accuracy: ± 3 MW 𝑍 𝑚𝑒𝑎𝑠 = vector of measured values
29 30

29 30

5
 09/11/2022

Example 1: WLS for 3-bus with 3 measurements system Example 1: WLS for 3-bus with 3 measurements system
-1

31 32

31 32

Example 1: WLS for 3-bus with 3 measurements system Example 2: WLS for 3-bus with 3 measurements system
Suppose the readings obtained are:
M12 = 62 MW M12 = 62 MW = 0.62 pu
M13 = 6 MW M13 = 6 MW = 0.06 pu
M32 = 37 MW M32 = 37 MW = 0.37 pu
Find the best estimate of the phase angles using WLS. Assume that
the three meters have the following.
Compare with the
case of using two Meters M12, M32 : 100 full-scale MW
: ± 3 MW accuracy
readings only.
Meter M13 : 100 full-scale MW
: ± 0.3 MW accuracy
33 34

33 34

Example 2: WLS for 3-bus with 3 measurements system Example 2: WLS for 3-bus with 3 measurements system
Solution
We will assume that the meter’s accuracy is being stated as equal to the
± 3. Then,
12 = 1.0 MW = 0.01 pu = 32
13 = 0.1 MW = 0.001 pu
The covariance matrix X

35 36

35 36

6
 09/11/2022

Example 2: WLS for 3-bus with 3 measurements system State Estimation of an AC Network
M12 = 62 MW Objective: minimize the sum of measurement residuals:
𝑁𝑚
M13 = 6 MW 𝑍𝑖𝑚𝑒𝑎𝑠 − 𝑓𝑖 (𝑥) 2
min 𝐽 𝑥 = ෍
M32 = 37 MW 𝜎𝑖2
𝑖=1
Measurements: MW, MVAR, MVA, amperes, transformer tap
Note
position, and voltage magnitude
• estimated flow on line 1–3 much States:
closer to the meter Reading. • voltage magnitude at each bus,
• estimated flow on lines 1–2 and • the phase angles at all buses except the reference bus,
3–2 are now further from the • the transformer taps
meter readings. Note: fi(x) functions will be nonlinear functions, except for a
37
voltage magnitude measurement 38

37 38

Sources of Error in State estimation What can phasor measurements do for a state estimator?
• Modeling Errors: wrong impedances, wrong • The PMUs do improve the state estimator accuracy
switch and breaker status in SCADA system through the following means:
• Data Errors: Wrong measurement direction ❖ The time accuracy of PMU measurements means that
specified in SCADA data base time differences now taken for granted in SCADA systems
are largely eliminated.
• Transducer Errors: Transducer magnetic amplifier ❖ Providing direct measurement of a voltage magnitude
or other electronics are not correctly calibrated and phase allows the state estimator to have additional
measurements for the same number of states
• Sampling Errors: not all measurements are taken
simultaneously
39 40

39 40

41

41