Example for Linear Programming

and Nonlinear Programming OPF

EE 521 Analysis of Power Systems
11-1-2012

Optimization Problem
Fixed problem:

f ( x) = 0

Optimization problem:
min f ( x)

What our world is

How to make our world

better

Optimal Power Flow

(Cost function)

u : decision variables (independent variables)

Generator voltage magnitude & real power

Voltage magnitude & angle at slack bus
Real power flow through dc lines
Phase angles across phase-shifting transformers
Turn ratios of tap-changing transformers
Admittances of variable reactors and switched capacitor banks
breaker positions by which the network can be reconfigured

Optimal Power Flow

z : exogenous variables (specified)

Real & reactive demands at load buses
Tie line flows
Admittance matrix
x : dependent variables

Real & reactive power at slack bus

Reactive power & angle at PV buses
Voltage magnitudes & angles at PQ buses

Optimal Power Flow

(Cost function)

Linear
Or
Nonlinear

How to form:
Cost function
Power flow equalities
Operating limit inequalities

Three ways to form the OPF problem

6-bus system

Linear Programming OPF(LPOPF)

Method 1

Objective Function Linear Cost Function

Fi ( Pi ) = si ( Pi Pi min ) + Fi ( Pi min ) min si Pi
i

Power Flow equality: Power balance equation

P1 + P2 + P3 = Pload + Plosses

Operating constraints:

P = B

AT B 1 P

P min P P max

=0
in

DC power flow

Linear Programming OPF(LPOPF)

6-bus system:

AT B 1 P

B 1

AT

P2
P
3
P4

P5
P6

independent variables
exogenous variables
(specified)

How to eliminate the exogenous

variables?
?

Pki = Gk i Pi

P2
P
3

Gk i Linear sensitivity coefficient

(Generation shift distribution factor, GSDF)

Linear Programming OPF(LPOPF)

6-bus system:

min
s.t.

12.4685P1 + 11.2887 P2 + 11.8333P3

P1 + P2 + P3 =
210MW
50 P1 200
37.5 P2 150
45 P3 180
AT B 1 P

Linear Programming OPF(LPOPF)

Result:

min
s.t.

12.4685P1 + 11.2887 P2 + 11.8333P3

P1 + P2 + P3 =
210MW
50 P1 200

Without line flow constraints:

P1* = 50, P2* = 115, P3* = 45

37.5 P2 150
45 P3 180
AT B 1 P

With line flow constraints:

P1 *= 79.8, P2* = 83.01, P3* = 47.19

Linear Programming OPF(LPOPF)

Method 2

Objective Function Piecewise Linearization

min
Fi ( P=
F
P
)
(
) + si1Pi1 + si 2 Pi 2 + si3 Pi3
i
i
i

0 Pik Pik+ , k =
1, 2,3
Pi = Pi min + Pi1 + Pi 2 + Pi 3

Linear Programming OPF(LPOPF)

Objective Function

Nonlinear Programming OPF(NLPOPF)

Method 3

Objective Function Quadratic cost function

P1* = 76.86,
P2* = 76.41
P3* = 56.73

Nonlinear Programming OPF(NLPOPF)

Method 4

Constraints:

Can we consider network losses?

P1 + P2 + +Pn = Pload + Plosses

Can we consider N-1 security?

Can we consider bus voltage constraints?

Further Information

Example 13c in Power Generation, Operation

and Control, A. J. Wood and B.F. Wollenberg,
Wiley, 2nd Edition, 1996.

Appendix: Kuhn-Tucker Condition

min f ( x)
s.t.g ( x) = 0
h( x ) 0

Appendix: Kuhn-Tucker Condition

Example:

min x12 2 x1 + x22 + 1

s.t.h1 ( x1 , x2 ) = x1 + x2 0
h2 ( x1 , x2 ) = x12 4 0

Kuhn-Tucker Condition:
(1)1 0, 2 0
(2)1h1 ( x1 , x2 ) = 1 ( x1 + x2 ) = 0
) 2 ( x12 4)
2 h2 ( x1 , x2=
= 0
(3)2 x1 2 + 1 + 22 x1 =
0
2 x2 + 1 =
0

Appendix: Kuhn-Tucker Condition

(1)1 0, 2 0
(2)1h1 ( x1 , x2 ) = 1 ( x1 + x2 ) = 0
= 0
2 h2 ( x1 , x2=
) 2 ( x12 4)
(3)2 x1 2 + 1 + 22 x1 =
0
2 x2 + 1 =
0

(1)(2)(3) only have 2 solutions :

x*1 1,=
x*2 0,=
=
*1 0,=
*2 0
1
1
x*1 =
, x*2 =
1, *2 0
, *1 ==
2
2

The first solution must be discarded because x*=(1,0) is

not feasible for the problem.
On the other hand, the second solution is feasible and is
the only solution of the optimal problem.