Scribd
Upload
38 views5 pages

Module 3: Frequency Control in A Power System Lecture 12a: Solution of Non-Linear Algebraic Equations

This document discusses solving nonlinear algebraic equations, which is necessary to compute the steady-state frequency of a power system through load flow analysis. It reviews fixed point iteration, a numerical technique to approximate solutions to nonlinear equations by guessing initial values and iteratively updating them. The method works if the initial guess is close to the solution, causing subsequent guesses to converge toward it. However, alternative methods like Newton-Raphson may converge faster for some equations. An example problem formulates the nonlinear load flow equations for a sample power system and sets up the fixed point iteration algorithm to solve for the unknown voltages and angles.

Uploaded by

Ashwani Rana
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
38 views5 pages

Module 3: Frequency Control in A Power System Lecture 12a: Solution of Non-Linear Algebraic Equations

This document discusses solving nonlinear algebraic equations, which is necessary to compute the steady-state frequency of a power system through load flow analysis. It reviews fixed point iteration, a numerical technique to approximate solutions to nonlinear equations by guessing initial values and iteratively updating them. The method works if the initial guess is close to the solution, causing subsequent guesses to converge toward it. However, alternative methods like Newton-Raphson may converge faster for some equations. An example problem formulates the nonlinear load flow equations for a sample power system and sets up the fixed point iteration algorithm to solve for the unknown voltages and angles.

Uploaded by

Ashwani Rana
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Module 3 : Frequency Control in a Power System Lecture 12a : Solution of non-linear algebraic equations

Non-linear algebraic equations and their solution In the next lecture, we will compute the steady state frequency of a power system, given the load characteristics. We shall see that in general, this will require us to carry out a "loadflow". A loadlfow involves the solution of a set of non-linear algebraic equations. Therefore, in this lecture we revise the basic methods to solve non-linear algebraic equations. We are aware that a transmission network in sinusoidal state state can be modelled by linear algebraic equations in the node voltage phasors( V) and the nodal current phasor injections ( I):

where Ybus is the bus admittance matrix. However, in power system studies, nodal injections are not specified as current phasors but as real and reactive power injections (nonlinear functions of V and I) , and/or voltage magnitudes of some nodes. We have also seen that real and reactive power can be a function of frequency. In such a situation, obtaining the steady state solution (i.e. node voltage phasors and frequency) will require us to solve a set of non-linear equations. Therefore we take a silight diversion from the main theme and review why and how we use numerical techniques for solving non-linear algebraic equations. Let us consider the "why" question first. If we wish to solve an equation of the form:

Perhaps, "simplifying" it will help us solve it ?

Perhaps, if we take the natrual logarithm of both sides we may be able to do something ?

But soon enough you will realize that we seem to be getting nowhere ! It is clear that some other way (guess work ?) may be required to get the solution. ixed Point Iteration Method Since we have some idea of how the exponential function behaves we can try to guess the solution. We know that:

and

and

We can guess that the solution for

should lie between 0.5 and 1.

However, this is a rough estimate. Surprisingly if we take an initial guess value :

and iterate as follows starting with k=0:

then x1=0.606, x2=0.545 , x3= 0.579, x4=0.5600, x5=0.571, x6=0.565, x7=0.568, x8=0.566, x9=0.567 .... We seem to be "converging" to a solution which satisfies the equation Why does the Fixed Point Iterative Method Work ? We can try to understand why we converge to the right solution by examining the behaviour of the iterative method near the solution. Suppose the correct solution to the equation is x = xs , i.e., !

Suppose the value of x at the kth iteration is near the solution xs and differs from it by a small amount Dxk , i.e.,

then:

which yields :

therefore if at k=0, x = xinit then,

Since:

Therefore as k tends to infinity, Dxk tends to zero. This means that if we start close enough to the solution, we will converge to the correct solution after many iterations, i.e., xk = xs if k is large. We say that the solution has converged if :

where e is the desired tolerance. The convergence is affected by the properties of the function at the solution point. You can check that if we wish to find out the solution of :

by the iterative algorithm:

the solution will diverge for any value of the initial guess which is not the true solution (x = 1). What happens if we formulate the iterative algorithm for the same equation as follows ?

ixed Point Iteration Method We now formally write the algorithm down. If we wish to find the solution for a set of n algebraic equations in n variables:

We first write the equations in the form (there is no unique way to to this!):

We then guess the initial values of all n variables and iterate as follows:

Unfortunately, this method does not converge always to the true solution for all nonlinear equations (as seen in the previous slide) and even if it does it sometimes converges very slowly. Some variations are proposed to improve convergence. For example, for the solution of , instead of using , one may first compute:

where a is a value which is between 0 and 1. It is found that this may speed up convergence for certain nonlinear equations. Alternative methods like Newton Raphson(N-R) can also be used for solving nonlinear equations. For an equation , N-R algorithm is :

where the partial derivative is Exercise:

evaluate d

at x=x k .

Can you write down the nonlinear algebraic equations corresponding to the real and reactive balance equations at each node for the system shown below ? The numerical values of the "known quantities" are indicated on the figure.

Can you formulate the iterative algorithm to find out the unknown quantities, i.e., the bus voltage phasors at buses 2 and 3, and the bus voltage phasor angle at bus 4, using the fixed point iterative method? Note that the bus voltage phasor at bus 1, voltage magnitude and real power injection at bus 4 is indicated on the figure.

Recap In this lecture you have learnt the following A recap on how to solve non-linear algebraic equations. Congratulations, you have finished Lecture 12a. To view the next lecture select it from the left hand side menu of the page.

You might also like