Assignment 2 ELL363

Ankit Mukherjee (2021EE30209), Harshit Gupta (2021EE30722),

Himanshu (2021EE30728), Kashish Kumawat (2021EE30739)
November 16, 2024

MCQs:
Q1:
For a 4-bus system using DC power flow, the initial state estimates (θ) are:
 
0
−0.02
θ= −0.03


−0.04

The power flow measurement vector (z) is:

 
0.5
 0.3 
z= −0.2


−0.1

The measurement Jacobian (H) is:

 
1 −1 0 0
0 1 −1 0 
H= 
0 0 1 −1
1 0 0 −1

Assumption: Measurement errors have a standard deviation of 0.02 per unit.

The residual error after one iteration is calculated as:

e = z − h, where h = H · θ

Options for Residual Error (e)

1. e = 0.014
2. e = 0.018
3. e = 0.020
4. e = 0.022

1
Soln:
State Estimates (θ):  
0
−0.02
θ=
−0.03


−0.04
These represent the voltage angles at each bus.
Measurement Vector (z):  
0.5
 0.3 
z= −0.2


−0.1
This is the actual power flow measured in the system.
Measurement Jacobian (H):
 
1 −1 0 0
0 1 −1 0 
H= 
0 0 1 −1
1 0 0 −1

The residual error (e) is defined as the difference between the measured values (z)
and the calculated values (h) based on the state estimates.

Now, the calculated power flow values (h) are given by:

h=H ·θ

Substitute H and θ:
   
1 −1 0 0 0
0 1 −1 0  −0.02
h= · 
0 0 1 −1 −0.03
1 0 0 −1 −0.04

Perform the matrix multiplication:

 
(1 · 0) + (−1 · −0.02) + (0 · −0.03) + (0 · −0.04)
(0 · 0) + (1 · −0.02) + (−1 · −0.03) + (0 · −0.04)
h= (0 · 0) + (0 · −0.02) + (1 · −0.03) + (−1 · −0.04)


(1 · 0) + (0 · −0.02) + (0 · −0.03) + (−1 · −0.04)

Simplify:  
0.02
0.01
h= 
0.01
0.04
Residual error is given by:
e=z−h

2
 Substitute z and h:    
0.5 0.02
 0.3  0.01
−0.2 − 0.01
e=   

−0.1 0.04
Perform the subtraction:
   
0.5 − 0.02 0.48
 0.3 − 0.01   0.29 
e=
−0.2 − 0.01 = −0.21
  

−0.1 − 0.04 −0.14

Now, the magnitude of the residual error (|e|) is given by:

v
u n
uX
|e| = t e2i
i=1

Substitute e:
p
|e| = (0.48)2 + (0.29)2 + (−0.21)2 + (−0.14)2

Calculate each term:

√
|e| = 0.2304 + 0.0841 + 0.0441 + 0.0196

Simplify: √
|e| = 0.3782
Final result:
|e| ≈ 0.615

Q2:
Problem: In a system with three thermal units, the optimal dispatch neglecting losses for
a 190 MW load is given. If the loss formula is applied, which of the following parameters
is critical to estimating incremental losses accurately?

1. Generation Shift Factors (GSFs)

2. Load Distribution Factors (LDFs)
3. Loss sensitivity coefficients
4. All of the above

Answer: To calculate incremental losses, the quadratic loss formula is used:

N X
X N
Ploss = Bij Pi Pj
i=1 j=1

where Bij are the loss coefficients.

Explanation:

3
 ˆ Generation Shift Factors (GSFs): GSFs quantify how redispatch affects line
flows.

ˆ Load Distribution Factors (LDFs): LDFs describe the redistribution of power

due to load changes.

ˆ Loss Sensitivity Coefficients: These are calculated as:

N
∂Ploss X
=2 Bij Pj
∂Pi j=1

Loss sensitivity coefficients refine how incremental changes in generation or load

affect losses.

All three—GSFs, LDFs, and loss sensitivities—are essential for accurate incremental
loss calculations in systems.
Correct Answer:
(4) All of the above.

Short answer question

Q3:
In contingency analysis, the maximum permissible load increase Pimax at bus i is expressed
as a function of GSFs, LODFs, and initial power flows. Derive the expression for Pimax
and discuss how it helps identify insecure lines under stressed conditions.

Soln:
Contingency analysis identifies insecure lines under stressed conditions by determining
the maximum permissible load increase at a bus before a line flow exceeds its limit. This is
done using Generation Shift Factors (GSFs), Line Outage Distribution Factors (LODFs),
and the initial power flows.

GSF (Generation Shift Factor): GSF quantifies how a change in generation at bus
i affects the flow on line k. Mathematically:

∆Fk = GSFk,i · ∆Pi

LODF (Line Outage Distribution Factor): LODF determines how the flow on line
l changes due to the outage of another line k. Mathematically:

∆Fl = LODFl,k · Fkpre

Initial Flow: The initial flow on line l is Flpre , and its limit is Flmax . Under stressed
conditions:
Flpost = Flpre + ∆Fl ≤ Flmax

4
Flow Constraint: For line l under contingency:
Flpost = Flpre + GSFl,i · Pimax + Effect of other line outages.
Assuming a single generator change and a single contingency:
Flpost = Flpre + GSFl,i · Pimax + LODFl,k · Fkpre

Flow Limit Condition: To ensure the line remains secure:

Flpost ≤ Flmax

Solve for Pimax : Rearranging:

Flmax − Flpre − LODFl,k · Fkpre
Pimax ≤
GSFl,i
Thus:
Flmax − Flpre − LODFl,k · Fkpre
Pimax =
GSFl,i

Insecure Line Detection: If Pimax is small, the line is close to its limit and is flagged
as insecure under contingency conditions.

Stress Analysis: This formula helps identify buses where load or generation increases
have a significant impact on line flows, allowing for preemptive corrective actions.
Consider a 3-bus system with the following parameters:
ˆ Line l: Flmax = 100 MW, Flpre = 80 MW
ˆ Line k: Fkpre = 50 MW
ˆ LODF: LODFl,k = 0.1
ˆ GSF: GSFl,i = 0.2
Using the formula:
100 − 80 − 0.1 · 50
Pimax =
0.2
100 − 80 − 5 15
Pimax = = = 75 MW
0.2 0.2
The maximum permissible load increase at bus i is 75 MW.

Q4:
Problem: A multi-area system consists of two areas, each with a governor response of
R1 = 0.01 pu, R2 = 0.02 pu.
A 100 MW load increase in Area 1 is managed by both areas. How would you integrate
Load Frequency Control (LFC), Economic Dispatch (ED), and Unit Commitment (UC)
to ensure frequency regulation while minimizing costs?
Considerations:
ˆ Tie-line constraints
ˆ Time-dependent Unit Commitment (UC) decisions

5
Soln:
The goal is to manage a 100 MW load increase in Area 1 using:

1. Load Frequency Control (LFC): for frequency regulation,

2. Economic Dispatch (ED): to minimize costs, and

3. Unit Commitment (UC): to determine which units should be online, considering

operational constraints and tie-line power flows.

LFC ensures frequency remains within acceptable bounds by redistributing power

between areas while maintaining balance between generation and demand.

ˆ The governor responses for the two areas are:

R1 = 0.01 pu, R2 = 0.02 pu.

ˆ The total equivalent governor response Req for the system is:

R1 R2
Req = .
R1 + R2

Substituting the given values:

0.01 · 0.02
Req = = 0.00667 pu.
0.01 + 0.02

ˆ The frequency deviation ∆f due to the 100 MW load change in Area 1 is:

∆Pload
∆f = .
Req

Substituting:
100
∆f = ≈ 15.0 Hz.
0.00667
ˆ Area 1 and Area 2 share the load change based on their individual governor re-
sponses:
R2 R1
∆P1 = ∆Pload , ∆P2 = ∆Pload .
R1 + R2 R1 + R2
Substituting:
0.02
∆P1 = · 100 = 66.67 MW,
0.01 + 0.02
0.01
∆P2 = · 100 = 33.33 MW.
0.01 + 0.02
Thus:

– Area 1 supplies 66.67 MW.

– Area 2 supplies 33.33 MW via the tie-line.

6
 Economic dispatch minimizes the total cost of generation while meeting demand and
respecting operational constraints.
Cost Functions: Let the cost functions of generators in Areas 1 and 2 be:

C1 (P1 ) = a1 + b1 P1 + c1 P12 , C2 (P2 ) = a2 + b2 P2 + c2 P22 .

Lagrangian Function:

L = C1 (P1 ) + C2 (P2 ) + λ (P1 + P2 − 100) .

Optimality Conditions:
∂L
1. ∂P1
= b1 + 2c1 P1 − λ = 0,
∂L
2. ∂P2
= b2 + 2c2 P2 − λ = 0,

3. P1 + P2 = 100, MW

Solving these equations provides the optimal generation values P1∗ and P2∗ .
Unit Commitment decides which generating units to turn on or off based on system
constraints and time-dependent operational decisions.
Constraints:

ˆ Generator Limits: Pmin,i ≤ Pi ≤ Pmax,i ,

ˆ Minimum Up/Down Time: Units must remain online or offline for a minimum
duration,

ˆ Tie-Line Capacity: Power flow through the tie-line must not exceed its thermal
limit:
Ptie = ∆P2 = 33.33 MW.

UC Decisions: Based on the demand forecast and reserve requirements, UC deter-

mines which units in Areas 1 and 2 should be online to supply 100 MW while respecting
constraints.

ˆ LFC: Immediate response to load change ensures frequency stability and power-
sharing between areas.

ˆ ED: Optimizes generation allocation post-LFC response to minimize costs.

ˆ UC: Ensures operational constraints and long-term reliability.

Final Result
ˆ Frequency deviation after LFC: ∆f = 15.0 Hz,

ˆ Power shared:
Area 1: 66.67 MW, Area 2: 33.33 MW,

ˆ ED optimally allocates generation based on cost functions,

ˆ UC ensures operational constraints and long-term reliability.

7
Analytical Question
Q5:
Consider a 4-bus power system with the following line reactances (in per unit):
ˆ X12 = 0.1

ˆ X23 = 0.15

ˆ X34 = 0.12

ˆ X14 = 0.2

Power flow measurements at the following locations with their respective values and
standard deviations (σ):
ˆ P12 = 0.8 pu, σ1 = 0.02

ˆ P23 = 0.5 pu, σ2 = 0.02

ˆ P34 = 0.3 pu, σ3 = 0.02

Assuming θ1 = 0 as reference:

a) Measurement Equation Formulation

Formulate the DC state estimation problem’s measurement equation z = Hx + e

b) State Estimation
Calculate the state estimate x̂ = (H T R−1 H)−1 H T R−1 z where R is the measurement
covariance matrix

c) Bad Data Analysis

Calculate the measurement residuals and determine if any measurement might be bad
data using the 3σ criterion

Solution
Given Data:
ˆ Line reactances (pu):

– X12 = 0.1
– X23 = 0.15
– X34 = 0.12
– X14 = 0.2

ˆ Measurements:

8
 – P12 = 0.8 pu, σ1 = 0.02
– P23 = 0.5 pu, σ2 = 0.02
– P34 = 0.3 pu, σ3 = 0.02

(a) Measurement Equation Formulation

State Vector (x):
x = [θ2 , θ3 , θ4 ]T (θ1 = 0 as reference)
Measurement Vector (z):

z = [P12 , P23 , P34 ]T = [0.8, 0.5, 0.3]T

H Matrix Formation:
For DC power flow:
θi − θj
Pij =
Xij
1) For P12 measurement:
θ1 − θ2 0 − θ2
P12 = =
X12 0.1
∂P12 1
=− = −10
∂θ2 0.1
∂P12
=0
∂θ3
∂P12
=0
∂θ4
2) For P23 measurement:
θ2 − θ3
P23 =
X23
∂P23 1
= ≈ 6.67
∂θ2 0.15
∂P23 1
=− ≈ −6.67
∂θ3 0.15
∂P23
=0
∂θ4
3) For P34 measurement:
θ3 − θ4
P34 =
X34
∂P34
=0
∂θ2
∂P34 1
= ≈ 8.33
∂θ3 0.12
∂P34 1
=− ≈ −8.33
∂θ4 0.12

9
 Therefore:  
−10.0 0.0 0.0
H =  6.67 −6.67 0.0 
0.0 8.33 −8.33
Measurement Error Covariance Matrix (R):
R = diag([σ12 , σ22 , σ32 ])
R = diag([0.022 , 0.022 , 0.022 ])
R = diag([0.0004, 0.0004, 0.0004])

(b) State Estimation Calculation

1) Calculate R−1 :
R−1 = diag([2500, 2500, 2500])
2) Calculate H T R−1 :
 
−25000 16675 0
H T R−1 = 0 −16675 20825 
0 0 −20825
3) Calculate H T R−1 H and solve (H T R−1 H)x̂ = H T R−1 z
Final State Estimate:
x̂ = [θ2 , θ3 , θ4 ]T = [−0.080, −0.155, −0.191]T radians

(c) Measurement Residuals and Bad Data Detection

Estimated Measurements (H x̂):
0 − (−0.080)
P12 est = = 0.800 pu
0.1
−0.080 − (−0.155)
P23 est = = 0.500 pu
0.15
−0.155 − (−0.191)
P34 est = = 0.300 pu
0.12
Measurement Residuals (r = z − H x̂):
r1 (P12 ) = 0.800 − 0.800 = 0.000 pu
r2 (P23 ) = 0.500 − 0.500 = 0.000 pu
r3 (P34 ) = 0.300 − 0.300 = 0.000 pu
Normalized Residuals: All normalized residuals = 0 < 3σ
Bad Data Detection: None of the measurements exceed the 3σ threshold, indicating
no bad data.
Validation:
Angle Differences:
θ1 − θ2 = 0.080 rad = 4.58◦
θ2 − θ3 = 0.075 rad = 4.30◦
θ3 − θ4 = 0.036 rad = 2.06◦

10
 Power Flow Check:

P12 = 0.800 pu (matches measurement of 0.8 pu)

P23 = 0.500 pu (matches measurement of 0.5 pu)
P34 = 0.300 pu (matches measurement of 0.3 pu)

Conclusion:

1. The state estimation successfully converged to a solution

2. All residuals are effectively zero, indicating perfect fit

3. No bad data detected in measurements

4. Angle differences are within reasonable ranges for power systems

5. Estimated states reproduce the measured power flows exactly

The solution is physically reasonable and mathematically consistent, with power flows
matching the measurements and angle differences within typical ranges for power systems.

11