OPTIMAL POWER FLOW

Debasish Choudhury
Assam State Electricity Grid Corporation Ltd.
Guwahati-781001
 OUTLINE

Introduction
OPF Formulation
Different Methodologies of OPF solution
Applications
Conclusion
References
 INTRODUCTION

The concept of OPF was first introduced by Carpentier in 1962 as an

extension of the conventional ELD problem.
An OPF problem schedules the power system controls to optimize the
specific objective function while satisfying the physical and operational
constraints of the electric network.
Basically an OPF is a static, constrained, non-linear optimization
problem.
Optimal power flow analysis is highly important as it helps provide
accurate calculations in the future expansion of power systems. It also
helps in determining the best type of operation of existing systems.
 OPF FORMULATION
oObjective Function:
o The most common OPF objective is the minimization of
generation costs, with or without consideration of system losses.

oVariables:
o Control Variables
o Dependent Variables

oConstraints:
o Equality Constraints
o Inequality Constraints
 OPF FORMULATION(Contd)
OPF is formulated mathematically as a general constrained optimization
problem.

Minimize f(u,x)

Subject to g(u,x)=0 (1)

h(u,x) 0 (2)

The objective function f(u,x) represents the systems goal which is a scalar
function. Vector functions g(u,x) and h(u,x) are the equality and
inequality constraints respectively.

Where u=the set of controllable quantities in the system

x=the set of dependent variables

 DIFFERENT METHODOLOGIES
The optimization techniques are mainly of two types:
Conventional Methods
1)Nonlinear Programming Method
2)Quadratic Programming Method
3)Newtons Method
4)Linear Programming Method
5)Interior Point Method
Artificial Intelligence Methods
1)Genetic Algorithm Method
2)Particle Swarm Optimization Method
 NLP Method
This is the earliest formulation category. In 1962, Carpentier first
introduced a generalized nonlinear programming formulation of the
economic dispatch problem, including voltage and other constraints.
The optimality conditions using Kuhn-Tucker method can be derived by
formulating the Lagrange function (L):

The conditions for an optimum for the points

1) L

0
x ,
0 0
( , ) 0
x

2) g ( x
0
) 0

3) h(x ) 0
0

0 0
h(x ) 0
4)

0
0
 Merits and Demerits
Merits:
o Capture system behavior accurately.
o Formulation is simple.

Demerits:
o The sufficiency conditions are difficult to verify.
o Slow in Convergence.
o Difficult to solve in presence of inequality
constraints.
 QP Method
Quadratic programming method is a special form of nonlinear
programming whose objective function is quadratic with linear
constraints. In 1973, Reid and Hasdorf presented a quadratic
programming method to solve QP problems.
0
Step1: Initial solutionx Hessian approximationB0 and penalty parameter r
> 0 are set up. and k = 0.
Step2: Quadratic programming problem is defined by objective function and
constrained condition.
Minimize
f ( x k ) d k d k L ( x k , k ) d k
T T 2

Subjects to:
ci (xk) c i ( x k ) T
d k
0

c ( x ) c i ( x k )
T
i k d k
0
Where

d k
x k 1
x k
 QP Method (Contd)
Step3: Penalty parameter is updated. Here, > 0.

Where is the Lagrangian Multiplier.

Step4: Linear search is done, step width is searched, and

Step5: Hessian approximation B is updated by BFGS algorithm,

T
T k
k
k 1 k
k
B s (s ) B
k (k ) k
y (y )
B B T
T
s B ( s ( k )) k
k k k
(y ) s
Where
 QP Method (Contd)
Step6: When convergence condition is fulfilled, this calculation is finished.
Otherwise k=k+1 and returns to step 2.
Merits:
o In this QP method, convergence is very fast.
o The method can solve both the load flow and economic dispatch
problems.
o The accuracy of QP method is much higher compared to other
established methods.
Demerits:
o Difficulties in obtaining solution of quadratic programming in large
dimension of approximating QP problems.
o Complexity and reliability of quadratic programming algorithms.
 NEWTONs Method
To speed up the convergence, in 1974, M.H. Rashed presented a method using
Lagrange multiplier and Newtons method, the method also introduced an
acceleration factor to compute the update controls.

Step1: Make initial guesses of vector and which inequality

constraints to enforce.

Step2: Create the Lagrangian given active inequality constraints.

Step3: Calculate the Hessian and gradient of the Lagrangian.

Step4: Solve the equation,

Step5: Calculate new z ,

 NEWTONs Method(Contd)
Step6: Check tolerance , , if not then go to step 3.
Step7: If correct inequalities are enforced, then problem completed, else go
to step2 by determining new set of inequalities to enforce using Lagrange
multipliers.
Merits:
o The method has the ability to converge fast.
o It can handle inequality constraints very well.
Demerits:
o It is not possible to develop practical OPF programs without employing
sparsity techniques.
o Sensitive to the initial conditions and they may even fail to converge due
to inappropriate initial conditions.
 LP Method
Linear Programming method treats problems having constraints and objective
functions formulated in linear form with non negative variables. In 1968, Wells
developed a linear programming approach to determine an economical
schedule. The cost objective and its constraints were linearized and solved
using the simplex method.

Step1: Initial power conditions.

Step2: Solve power flow equations(DC or AC power flow).

Step3: Create linear objective function.

Step4: Obtain linearized constraint sensitivity coefficients.

Step5: Set up and solve LP for new control variable settings

Step6: Check for convergence test, if not then goto step1 by adjusting one or more
control variables else stop the problem.
 LP Method(Contd)
Merits:
o The LP method easily handles Non linearity constraints.
o The LP solution is completely reliable.
o It has the ability to detect infeasible solution.
o The LP solution can be very fast.
o Complete computational reliability and very high speed enables it,
suitable for real time or steady mode purposes.
Demerits:
o It suffers lack of accuracy.
 IP Method
Karmarkar proposed a new method in 1984 for solving large-scale linear
programming problems very efficiently. It is known as an interior
method since it finds improved search directions strictly in the interior of
the feasible space.
The barrier function is in the form;

After introducing the barrier function, we can write the modified OPF
formulation:

The Lagrange function for this problem is:

 IP Method(Contd)
The KKT conditions are

These nonlinear equations are then solved iteratively by

Newtons method, and the value is adjusted toward zero.
 IP Method(Contd)

Merits:
o The Interior Point Method can solve a large scale linear
programming problem by moving through the interior, rather
than the boundary as in the simplex method, of the feasible
region to find an optimal solution.
o The Interior Point Method is preferably adapted to OPF due
to its reliability, speed and accuracy.
o IP provides user interaction in the selection of constraints.

Demerits:
o Limitation due to starting and terminating conditions.
o Infeasible solution if step size is chosen improperly.
 GA Method
GAs search within a population of points, not a single point. Therefore
GAs can provide a globally optimal solution. GAs use only objective
function information, not derivatives or other auxiliary knowledge.

Step-by-step algorithm for GA Method:

Step1: Read the database.

Step2:Assume suitably population size(pop_size) , maximum no. of

generations or population(gen_max).

Step 3: Set valid number of population counter. Pop_vn=0.

Step 4: Randomly generate the chromosomes.

Step 5: Run the power flow using the Newton-Raphson method.

Step 6: Check all inequality constraints, if any of the limits is violated, go to

step 4.
 GA Method(Contd)
Step 7: If all the constraints are satisfied, increment pop_vn by 1. If pop_vn
less than than or equal to pop_size, go to step 4, otherwise go to next
step.

Step8: Calculate and store the total cost of generation corresponding to each
valid generation pattern of chromosome.

Step 9: Find and store minimum cost among all valid indivisual parents and
corresponding generation pattern.

Step 10: Check if random no. (crossover rate) for i=1 to pop_size. Select the
ith chromosome. Apply the crossover operator to that indivisual.

Step 11: Run power flow using Newton-Raphson method and Check system
constrains as in step-6.
 GA Method(Contd)
Step 12: If all the constraints are satisfied, the individual of the new
population becomes valid otherwise it becomes invalid.

Step 13: Apply the mutation operator to the calculated generation patterns.

Step 14: Run power flow using Newton-Raphson and check all the
constraints.

Step 15: If all the constraints are satisfied go to next step otherwise go to
step 4.

Step 17: Calculate the total cost of all valid patterns.

Step 18: Find the optimum solution of among all.
 GA Method(Contd)
Merits:
o GAs can handle the Integer or discrete variables.
o GAs has the potential to find solutions in many different areas of the search
space simultaneously there by multiple objectives can be achieved in single
run.
o GAs are adaptable to change, ability to generate large number of solutions
and rapid convergence.
o GAs can be easily coded to work on parallel computers.
Demerits:
o The execution time and the quality of the solution, deteriorate with the
increase of the chromosome length, i.e., the OPF problem size.
o If the size of the power system is increasing, the GA approach can produce
more in feasible springs which may lead to wastage of computational efforts.
 PSO Method
Particle swarm optimization (PSO) is a population based stochastic
optimization technique inspired by social behavior of bird flocking or
fish schooling. In PSO, the search for an optimal solution is conducted
using a population of particles, each of which represents a candidate
solution to the optimization problem.
Step 1: Each member is called a particle, and each particle (i-th particle) is
represented by d-dimensional vector and described as

Step 2: The set of n particles in the swarm are called population and
described as pop=

Step 3: The best previous position for each particle (the positions giving the
best fitness value) is called particle best and described as
 PSO Method(Contd)
Step 4: The best position among all of the particle best position achieved
so far is called global best and described as

Step 5: The rate of position change for each particle is called the particle
velocity and it is described as

Step 6: At iteration k the velocity for d-dimension of ith particle is

updated by:

Step 7: Check the optimum case, if not then update individuals and go to
step 2.
 PSO Method(Contd)
Merits:
o These methods are simple concept, easy implementation, relative
robustness to control parameters and computational efficiency.
o The prominent merit of PSO is its fast convergence speed.
o PSO algorithm can be realized simply for less parameter adjusting.
o PSO has the flexibility to control the balance between the global and
local exploration of the search space.
Demerits:
o Real coding of these variables represents limitation of PSO methods as
simple round-off calculations may lead to significant errors.
o Slow convergence in refined search stage (weak local search ability).
 APPLICATIONS
Security Constrained Economic Dispatch

Preventive and Corrective Rescheduling

Reactive Power Planning and Voltage Control

Power Wheeling and Wheeling Loss calculation

Pricing of Real and Reactive Power

 CONCLUSION

In general, key developments in the OPF formulation have

been accompanied by developments in solution techniques.
Moreover, the application of new solution techniques has
paralleled increases in computing ability such that recent
OPF algorithms are much more computationally intensive
than the original gradient methods.
 REFERENCES
1. J. A. Momoh, M. E. El-Hawary and R. Adapa, "A review of selected optimal
power flow literature to 1993. Part-I: Non-linear and quadratic programming
approaches", IEEE Transactions on Power Systems, Feb.'99, pp. 96-104.
2. J. A. Momoh, M. E. El-Hawary and R. Adapa, "A review of selected optimal
power low literature to 1993. Part-II: Newton, linear programming and interior
point methods",IEEE Transactions on Power Systems, Feb.'99, pp. 105-111.
3. M. Huneault and F. D. Galiana, "A survey of the optimal power flow literature", I
IEEE Transactions on Power Systems, May '91, pp. 762-770.
4. J. Carpentier, "Contribution e I'etude do Dispatching Economique", Bulletin
Society Francaise Electriciens, August 1962.
5. A. J. Wood and B. F. Wollenberg, Power Generation. Operation and Control, 2nd
Edition, John Wiley and Sons, Inc. 1996.
6. H. W. Dommel and W. F. Tinney, "Optimal power flow solutions",
IEEE Transactions on Power Apparatus and Systems, October 1968, pp.1866-
1876.
THANK YOU