Réseaux de transport et de

distribution d’énergie électrique:

Power Flow

ALI MEHRIZI-SANI
ASSOCIATE PROFESSOR
Resilient Renewable Energy Grid Adaptation Laboratory (REGAL)

P
POOW
WEER
R A
ANND
D E
ENNE
ERRG
GYY C
CEEN
NTTE
ER
ESIGELEC R

Rouen, France • November 2023

Power Flow Formulation and
Solution
Power flow studies are at the heart of
power system studies; power flow
determines, based on the given
generation and load scenarios and the
network topology, what the line
currents and the bus voltages will be.
Topics
System operators continually run power
flow studies to ensure the power system
remains stable and currents and voltages
remain within prescribed ranges to avoid
system blackouts, outages, and unsafe
operation conditions.
• Review of standard KVL and KCL
equations
• Conversion of equation to power flow
(V-PQ) equations based on the typical
knowns (loads and generation) and
unknowns (voltages)
• Solution techniques
• Per-unit system
Power Flow
• The most common power system analysis tool is the power flow (also
known sometimes as the load flow)
– Power flow determines how the power flows in a network
– It also determines all bus voltages and all currents
– Because of constant power models, power flow is a nonlinear analysis technique
– Power flow is a steady-state analysis tool
– It can be used as a tool for planning the location of new generation, including wind.
Simplified Power System Modeling
• Balanced three-phase systems can be analyzed using per phase analysis

• A “per unit” normalization is used to simplify the analysis of systems

with different voltage levels (that is, when transformers are present)

• We need models for the different system devices:

– Transformers and transmission lines
• Normally modeled as their series impedance
– Loads
– Generators
Load Models
• The ultimate goal is to supply loads with electricity at constant
frequency and voltage

• Electrical characteristics of individual loads matter, but usually they can

only be estimated
– Actual loads are constantly changing, consisting of a large number of individual devices
– We have only limited network observability of load characteristics

• Therefore, aggregate models are typically used for analysis

• Two common models (ZIP) are

– constant power: Si = Pi + jQi
– constant impedance: Si = |V|2 / Zi
Generator Models
• Engineering models depend on application.

• Generators are usually synchronous machines.

• For generators we will use two different models:

– A steady-state model, treating the generator as a constant power source operating at a
fixed voltage; this model will be used for power flow and economic analysis
– Other models include treating as constant real power with a fixed power factor.
Power Flow Analysis
• We now have the necessary models to start to develop the power system
analysis tools.
Bus Admittance Matrix or Ybus (from nodal analysis)
• First step in solving the power flow is to create what is known as the bus
admittance matrix, often called the Ybus.

• The Ybus gives the relationships between all the bus current injections, I,
and all the bus voltages, V,
I = Ybus V

• The Ybus is developed by applying KCL at each bus in the system to relate
the bus current injections, the bus voltages, and the branch impedances
and admittances
Ybus Example
• Determine the bus admittance matrix for the network shown below,
assuming the current injection at each bus i is Ii = IGi - IDi where IGi is the
current injection into the bus from the generator and IDi is the current
flowing into the load
Ybus General Form
• The diagonal terms, Yii, are the self admittance terms, equal to the sum
of the admittances of all devices incident to bus i.

• The off-diagonal terms, Yij, are equal to the negative of the sum of the
admittances joining the two buses.

• With large systems Ybus is a sparse matrix (that is, most entries are zero)
Power Flow Analysis
• When analyzing power systems we know neither the complex bus
voltages nor the complex current injections

• Rather, we know the complex power being consumed by the load, and the
power being injected by the generators plus their voltage magnitudes

• Therefore we can not directly use the Ybus equations, but must use the
power balance equations
Power Balance Equations

From KCL we know at each bus i in an n bus system

the current injection, I i , must be equal to the current
that flows into the network
n
Ii I Gi - I Di å Iik
k 1
Since I = Ybus V we also know
n
Ii I Gi - I Di å YikVk
k 1
*
The network power injection is then Si Vi I i
Power Balance Equations, cont’d

*
æ n ö n
Si Vi I i* Vi ç å YikVk ÷ Vi å Yik*Vk*
èk 1 ø k 1
This is an equation with complex numbers.
Sometimes we would like an equivalent set of real
power equations. These can be derived by defining
Yik  Gik jBik
Vi  Vi e jqi Vi Ðq i
q ik  q i - q k
Recall e jq cosq j sin q
Real Power Balance Equations

n n
jq ik
Si Pi jQi Vi å Yik*Vk* å i k
V V e (Gik - jBik )
k 1 k 1
n
å Vi Vk (cosq ik j sin q ik )(Gik - jBik )
k 1
Resolving into the real and imaginary parts
n
Pi å Vi Vk (Gik cosqik Bik sin q ik ) PGi - PDi
k 1
n
Qi å Vi Vk (Gik sinqik - Bik cosqik ) QGi - QDi
k 1
Power Flow Slack Bus
• We can not arbitrarily specify S at all buses because total generation
must equal total load + total losses

• We also need an angle reference bus.

• To solve these problems we define one bus as the "slack" bus. This bus
has a fixed voltage magnitude and angle, and a varying real/reactive
power injection.
Bus Types in a Power Flow Problem
• PQ: A load bus where we know the real and reactive demand
(consumption) of power.

• PV: A generator bus where the real power is scheduled (known) and
voltage is regulated (also known)

• Vdelta: Slack bus

• What if a bus has both a load and a generator?

Nonlinear Systems: Multiple Solutions or No Solution
• Example 1: x2 - 2 = 0 has solutions x =E 1.414…
• Example 2: x2 + 2 = 0 has no real solution

x2
f(x) = x2 - 2 f(x) = x2 + 2
f(x) = = 0

↑
two solutions where f(x) = 0 no solution f(x) = 0
Multiple Solution Example
• The dc system shown below has two solutions:

where the 18 W
load is a resistive
load

The equation we're solving is

What is the
2 æ 9 volts ö
2
maximum
I RLoad ç ÷ RLoad 18 watts PLoad?
è 1W +R Load ø
One solution is R Load 2W
Other solution is R Load 0.5W
Newton-Raphson Method (scalar)

General form of problem: Find an x such that

f ( xˆ ) 0
1. For each guess of xˆ , x ( v ) , define
Dx ( v ) xˆ - x ( v )
2. Represent f ( xˆ ) by a Taylor series about f ( x )
(v)
(v) df ( x ) (v)
f ( xˆ ) f (x ) Dx
dx
1 d 2 f ( x( v ) ) (v) 2
Dx higher order terms
2 dx 2
Newton-Raphson Method, cont’d

3. Approximate f ( xˆ ) by neglecting all terms

except the first two
(v )
(v) df ( x ) (v)
f ( xˆ ) 0 » f (x ) Dx
dx
4. Use this linear approximation to solve for Dx ( v )
-1
(v) é df ( x ) ù
(v)
(v)
Dx -ê ú f ( x )
ë dx û
5. Solve for a new estimate of x̂
x(v 1)
x(v) Dx ( v )
Newton-Raphson Example

Use Newton-Raphson to solve f ( x) x2 - 2 0

The equation we must iteratively solve is
-1
(v) é df ( x ) ù(v)
(v)
Dx -ê ú f ( x )
ë dx û
é 1 ù (v) 2
Dx ( v ) - ê ( v ) ú (( x ) - 2)
ë2x û
x(v 1)
x(v) Dx ( v )
( v 1) (v) é 1 ù (v) 2
x x - ê ( v ) ú (( x ) - 2)
ë2x û
Newton-Raphson Example, cont’d

( v 1) (v) é 1 ù (v) 2
x x - ê ( v ) ú (( x ) - 2)
ë2x û
Guess x (0) 1. Iteratively solving we get
v x(v) f ( x(v) ) Dx ( v )
0 1 -1 0.5
1 1.5 0.25 -0.08333
2 1.41667 6.953 ´ 10-3 -2.454 ´ 10-3
-6
3 1.41422 6.024 ´ 10
Newton-Raphson Comments
• When close to the solution the error decreases quite quickly -- method
has quadratic convergence
• f(x(v)) is known as the mismatch, which we would like to drive to zero
• Stopping criteria is ½f(x(v)) ½ < e
• Results are dependent upon the initial guess. What if we had guessed
x(0) = 0, or x (0) = -1?
• A solution’s region of attraction (ROA) is the set of initial guesses that
converge to the particular solution. The ROA is often hard to determine
Multi-Variable Newton-Raphson

Next we generalize to the case where x is an n-

dimension vector, and f (x) is an n-dimension function
é x1 ù é f1 (x) ù
êx ú ê f ( x) ú
x ê 2ú f ( x) ê 2 ú
êú ê  ú
êx ú ê f ( x) ú
ë nû ë n û
Again define the solution xˆ so f ( xˆ ) 0 and
Dx xˆ - x
Multi-Variable Case, cont’d

The Taylor series expansion is written for each fi (x)

¶f1 (x) ¶f1 ( x)
f1 (xˆ ) f1 ( x) Dx1 Dx2 
¶x1 ¶x2
¶f1 (x)
Dxn higher order terms
¶xn

¶f n (x) ¶f n ( x)
f n (xˆ ) f n ( x) Dx1 Dx2 
¶x1 ¶x2
¶f n (x)
Dxn higher order terms
¶xn
Multi-Variable Case, cont’d

This can be written more compactly in matrix form

é ¶f1 (x) ¶f1 (x)

¶f1 (x) ù
ê ¶x ¶x2 ¶xn ú
1
é f1 (x) ùê ú é Dx1 ù
ê f ( x) úê ¶f 2 (x) ¶f 2 (x) ¶f 2 (x) ú ê ú
ê ¶x1
 D x
f (xˆ ) ê 2 ú ¶x2 ¶xn ú ê 2 ú
ê  ú ê úê  ú
ê f ( x) úê     úê ú
ë n û Dx
ê ¶f (x) ¶f n (x) ¶f n (x) ú ë n û
ê n  ú
êë ¶x1 ¶x2 ¶xn úû
higher order terms
Jacobian Matrix

The n by n matrix of partial derivatives is known

as the Jacobian matrix, J (x)
é ¶f1 (x) ¶f1 (x)

¶f1 ( x) ù
ê ¶x ¶x2 ¶xn ú
1
ê ú
ê ¶f 2 (x) ¶f 2 (x)

¶f 2 ( x) ú
J (x) ê ¶x1 ¶x2 ¶xn ú
ê ú
ê     ú
ê ¶f (x) ¶f n (x) ¶f n ( x) ú
ê n  ú
êë ¶x1 ¶x2 ¶xn úû
Multi-Variable N-R Procedure

Derivation of N-R method is similar to the scalar case

f (xˆ ) f (x) J (x)Dx higher order terms
f (xˆ ) 0 » f (x) J (x)Dx
Dx » -J (x) -1f (x)
( v 1) (v) (v)
x x Dx
x(v 1)
x( v ) - J (x( v ) ) -1 f (x( v ) )
Iterate until f (x( v ) ) e
Multi-Variable Example

é x1 ù
Solve for x = ê ú such that f (x) 0 where
ëx2 û
f1 (x) 2 x12 x22 - 8 0
f 2 ( x) x12 - x22 x1x2 - 4 0
First symbolically determine the Jacobian
é ¶f1 (x) ¶f1 ( x) ù
ê ¶x ¶x2 ú
1
J (x) = ê ú
ê ¶f 2 (x) ¶f 2 ( x) ú
êë ¶x1 ¶x2 úû
Multi-variable Example, cont’d

é 4 x1 2 x2 ù
J (x) = ê
ë 2 x1 x2 x1 - 2 x2 úû
Then
-1
é Dx1 ù é 4 x1 2 x2 ù é f1 (x) ù
ê Dx ú -ê
ë 2û ë 2 x1 x2 x1 - 2 x2 úû êë f 2 (x) úû
é1ù
Arbitrarily guess x(0) ê1ú
ëû
-1
(1) é1ù é 4 2 ù é -5ù é 2.1ù
x ê1ú - ê 3 -1ú ê -3ú ê1.3ú
ëû ë û ë û ë û
Multi-variable Example, cont’d

-1
(2) é 2.1ù é8.40 2.60 ù é 2.51ù é1.8284 ù
x ê1.3 ú - ê5.50 -0.50 ú ê1.45 ú ê1.2122 ú
ë û ë û ë û ë û
Each iteration we check f (x) to see if it is below our
specified tolerance e
(2) é0.1556 ù
f (x ) ê0.0900 ú
ë û
If e = 0.2 then we would be done. Otherwise we'd
continue iterating.
Newton-Raphson Power Flow

In the Newton-Raphson power flow we use Newton's

method to determine the voltage magnitude and angle
at each bus in the power system.
We need to solve the power balance equations
n
Pi å Vi Vk (Gik cosq ik Bik sin q ik ) PGi - PDi
k 1
n
Qi å Vi Vk (Gik sin q ik - Bik cos q ik ) QGi - QDi
k 1
Power Flow Variables

Assume the slack bus is the first bus (with a fixed

voltage angle/magnitude). We then need to determine
the voltage angle/magnitude at the other buses.
é q2 ù é P2 (x) - PG 2 PD 2 ù
ê  ú ê  ú
ê ú ê ú
ê qn ú ê Pn (x) - PGn PDn ú
x êV ú f ( x) êQ ( x) - Q ú
Q
ê 2ú ê 2 G2 D2 ú
ê  ú ê  ú
ê ú ê ú
ë Vn û ë Qn (x) - QGn QDn û
N-R Power Flow Solution

The power flow is solved using the same procedure

discussed last time:
Set v 0; make an initial guess of x, x( v )
While f (x( v ) ) e Do
( v 1) (v) ( v ) -1 (v)
x x - J (x ) f (x )
v v 1
End While
Power Flow Jacobian Matrix

The most difficult part of the algorithm is determining

and inverting the n by n Jacobian matrix, J (x)
é ¶f1 (x) ¶f1 (x)

¶f1 ( x) ù
ê ¶x ¶x2 ¶xn ú
1
ê ú
ê ¶f 2 (x) ¶f 2 (x)

¶f 2 ( x) ú
J (x) ê ¶x1 ¶x2 ¶xn ú
ê ú
ê     ú
ê ¶f (x) ¶f n (x) ¶f n ( x) ú
ê n  ú
êë ¶x1 ¶x2 ¶xn ûú
Power Flow Jacobian Matrix, cont’d

Jacobian elements are calculated by differentiating

each function, fi (x), with respect to each variable.
For example, if fi (x) is the bus i real power equation
n
fi ( x) å Vi Vk (Gik cosq ik Bik sin q ik ) - PGi PDi
k 1
n
¶fi ( x)
¶q i
å Vi Vk (-Gik sin q ik Bik cosq ik )
k 1
k ¹i

¶fi ( x)
Vi V j (Gik sin q ik - Bik cosq ik ) ( j ¹ i )
¶q j
 ->
&
j

+- (a)
-
S
-
&P,
G
x =
[i] -

4x2 :
types of expressions

INVP=(Ginos) +Bit
ii

i
Two Bus Newton-Raphson Example
• For the two bus power system shown below, use the Newton-Raphson
power flow to determine the voltage magnitude and angle at bus two.
Assume that bus one is the slack and SBase = 100 MVA.

Line Z = 0.1j

One 1.000 pu Two 1.000 pu

0 MW 200 MW
0 MVR 100 MVR

éq2 ù é - j10 j10 ù

x êV ú Ybus ê j10 - j10 ú
ë 2û ë û
Two Bus Example, cont’d

General power balance equations

n
Pi å Vi Vk (Gik cosqik Bik sin qik ) PGi - PDi
k 1
n
Qi å Vi Vk (Gik sin qik - Bik cos qik ) QGi - QDi
k 1
Bus two power balance equations
V2 V1 (10sin q 2 ) 2.0 0
2
V2 V1 ( -10cos q 2 ) V2 (10) 1.0 0
Two Bus Example, cont’d

P2 (x) V2 (10sin q 2 ) 2.0 0

2
Q2 (x) V2 (-10cosq 2 ) V2 (10) 1.0 0
Now calculate the power flow Jacobian
é ¶P2 (x) ¶P2 (x) ù
ê ¶q ¶V 2 ú
2
J ( x) ê ú
ê ¶Q 2 (x) ¶Q 2 ( x) ú
ê ¶q ¶ V 2 úû
ë 2

é10 V2 cosq 2 10sin q 2 ù

ê10 V sin q -10cosq 2 20 V2 úû
ë 2 2
Two Bus Example, First Iteration

(0) é0 ù
Set v 0, guess x ê1 ú
ë û
Calculate

(0)
é V2 (10sin q 2 ) 2.0 ù é 2.0 ù
f(x ) ê 2 ú ê1.0 ú
êë V2 (-10 cosq 2 ) V2 (10) 1.0 úû ë û
(0) é10 V2 cosq 2 10sin q 2 ù é10 0 ù
J (x ) ê10 V sin q ú ê 0 10 ú
ë 2 2 - 10 cos q 2 20 V2û ë û
-1
(1) é0 ù é10 0 ù é 2.0 ù é -0.2 ù
Solve x ê1 ú - ê 0 10 ú ê1.0 ú ê 0.9 ú
ë û ë û ë û ë û
Two Bus Example, Next Iterations

(1) é 0.9 (10sin(-0.2)) 2.0 ù é0.212 ù

f(x ) ê 2 ú ê0.279 ú
ë0.9(-10 cos(-0.2)) 0.9 ´ 10 1.0 û ë û
(1) é 8.82 -1.986 ù
J (x ) ê ú
ë -1.788 8.199 û
-1
(2) é -0.2 ù é 8.82 -1.986 ù é0.212 ù é -0.233ù
x ê 0.9 ú - ê -1.788 8.199 ú ê0.279 ú ê 0.8586 ú
ë û ë û ë û ë û
(2) é 0.0145ù (3) é -0.236 ù
f(x ) ê ú x ê 0.8554 ú
ë 0.0190 û ë û
(3) é0.0000906 ù
f(x ) ê 0.0001175ú Done! V2 0.8554Ð - 13.52
ë û
Two Bus Solved Values
• Once the voltage angle and magnitude at bus 2 are known we can
calculate all the other system values, such as the line flows and the
generator reactive power output

200.0 MW -200.0 MW
168.3 MVR Line Z = 0.1j -100.0 MVR

One 1.000 pu Two 0.855 pu -13.522 Deg

200.0 MW 200 MW
168.3 MVR 100 MVR
Three Bus PV Case Example

For this three bus case we have

é q2 ù é P2 (x) - PG 2 PD 2 ù
x êq ú f ( x) ê P ( x) - P P ú 0
ê 3ú ê 3 G3 D3 ú
êë V2 úû êë Q2 (x) QD 2 úû
Line Z = 0.1j

0.941 pu
One 1.000 pu Two -7.469 Deg

170.0 MW 200 MW
68.2 MVR 100 MVR
Line Z = 0.1j Line Z = 0.1j

Three 1.000 pu

30 MW
63 MVR
Solving Large Power Systems
• The most difficult computational task is inverting the Jacobian matrix
– Inverting a full matrix is an order n3 operation, meaning the amount of computation
increases with the cube of the size
– This amount of computation can be decreased substantially by recognizing that since
the Ybus is a sparse matrix, the Jacobian is also a sparse matrix
– Using sparse matrix methods results in a computational order of about n1.5.
– This is a substantial savings when solving systems with tens of thousands of buses
40 Bus Power Flow Example

Eagle
Island Electric Company (IEC)
1 .0 1 3 pu
13%
1 .0 1 8 pu
1 .0 2 0 pu 44% Robin
39% 38%
Pheasant 9 5 MW Flamingo 8 3 M var
5 0 5 MW
0 .9 9 2 pu

System Losses: 35.22 MW

1 1 0 MW
2 3 M var 8 7 MW
3 0 0 MW 1 9 M var 0 .9 9 9 pu 3 0 M var
24% 2 0 0 MW 13% 2%
1 .0 1 0 pu 6 0 M var

1 5 0 MW
7 5 MW
1 5 M var
Canary 1 .0 0 6 pu
1 9 8 MW
5% 1 6 0 MW
3 5 M var
1 .0 1 8 pu 7 0 M var 3 5 M var
19% 46% 5 6 M var
3 5 0 MW 0 .9 9 3 pu
1 .0 1 5 pu

Dove
7 8 M var 1 1 2 MW
Woodpecker
8%
28%
16% 17% 9%
4 0 M var
Rook 30%

1 .0 1 5 pu 1 .0 2 3 pu 1 .0 1 2 pu 1 1 5 MW
1 .0 1 8 pu 2 5 M var

Cardinal
12% 17%
24%
Parrot 1 .0 1 5 pu 35%
5 7 M var 1 .0 1 9 pu 2 6 8 MW 0 .9 9 1 pu
1 3 0 MW
3 0 M var Turkey 14% 1 2 8 M var
29%

Sparrow
1 .0 1 0 pu
10% 7%
8% 30% 1 .0 0 5 pu
1 .0 1 4 pu

Bluebird
0 .9 9 3 pu
6%
14% 45%
30% 13%
1 7 5 MW
Ostrich 1 .0 1 5 pu
0 .9 9 6 pu 18%
4 0 M var
0 .9 9 6 pu 1 3 2 MW
28%
Mallard 1 5 M var
7%
1 5 0 MW
3 9 M var 23%
5 1 M var
22%

6 0 MW
1 .0 0 2 pu Owl 0 .9 9 3 pu 27%
1 .0 1 4 pu
1 5 M var

Hawk
7 0 M var 1 3 0 MW

Piper
36%
4 5 M var 10%
0 .0 0 0 pu 2%
1 6 5 MW

Finch Oriole 35%

NewWind
3% 3 0 M var 0 .9 8 9 pu
7% 1 .0 0 1 pu 1 .0 1 5 pu
0 .9 9 4 pu
22% sl a ck
16%

Crow 0 .9 9 7 pu 45% 11% 1 4 0 MW 3 9 6 MW

0 MW 3%
1 3 5 MW
19%
1 2 8 MW
3 2 M var 1 7 6 MW
1 5 M var 33%
2 0 M var 0 .9 8 8 pu 2 8 M var 57%
5 5 MW
1 5 M var
6%
Lark 0 .9 9 6 pu
Heron 47%

Finch 0 .9 9 9 pu
22%
1 6 1 MW
53% 2 1 M var
48%
1 4 0 MW
2 0 M var
Condor
47% 1 .0 2 5 pu

Peacock 1 .0 0 4 pu 1 .0 1 0 pu

Hen
9 0 0 MW
8 8 MW 4% 1 1 2 0 MW
78% 78%
1 1 M var

1 .0 2 0 pu
Good Power System Operation
• Good power system operation requires that there be no reliability
violations for either the current condition or in the event of statistically
likely contingencies
• Reliability requires as a minimum that there be no transmission line/transformer limit
violations and that bus voltages be within acceptable limits (perhaps 0.95 to 1.08)
• Example contingencies are the loss of any single device.
• This is known as n-1 reliability.

• North American Electric Reliability Corporation has legal authority to

enforce reliability standards (and there are now lots of them).