Module 5 Contingency Analysis

5.1 Introduction and concept of linear sensitivity factors

In Chapter 2, we have discussed about the methods for analyzing the power system performance at a particular operating point. However, for practical system operation, apart from ensuring the satisfactory operation of the system at a particular operating condition, it is also equally important to make sure that the system operates with adequate level of security. Broadly, the term security implies the ability of the system to operate within system constraints (on bus voltage magnitudes, current and power ow over the lines) in the event of outage (contingency) of any component (generator or transmission line). Now, if the system is operating at high loading (light loading) conditions, then the post-contingency system condition would be highly stressed (lightly stressed). Therefore, the post contingency values of dierent quantities (voltages, current/power ow) depend on the present operating condition. In case the post-outage (post contingency) does not involve any violation of any operating constraints, the system is said to be operating securely. Otherwise, the system is said to enter an emergency operating condition. Therefore, for detecting the possibility of appearance of emergency operating conditions, analysis of the post-contingency scenario (henceforth termed as contingency analysis) of the system needs to be carried out. Now, in any practical sized power system, there is a very large number of elements. Hence, for carrying out contingency analysis, outages of all these elements (preferably) need to be carried out one-by-one corresponding to any particular operating condition. However, in any power system, the operating point of the system changes quite frequently with change is loading/generating conditions. With the change in system operating conditions, the contingency analysis exercise needs to be carried out again at the new operating point. Thus, for proper monitoring of system security, a large number of outage cases need to be simulated repeatedly over a short span of time. Ideally, these outage cases should be studied with the help of full AC load ow solutions. However, analysis of thousands of outage cases with full AC power ow technique will involve a signicant amount of computation time and as a result, it might not be possible to complete this entire exercise before the new operating condition emerges. Therefore, instead of using full non-linear AC power ow analysis, approximate, but much faster techniques based on linear sensitivity factors are used to estimate the post contingency values of dierent quantities of interest. The basic concept of sensitivity factors 226

is described below. Essestially, the linear sensitivity factors approximately estimate the changes in dierent line ows for any particular outage condition without the need of full AC power ow solution. Basically, there are two types of sensitivity factors and these are: a. Generation outage sensitivity factor (GOSF) b. Line outage sensitivity factor (LOSF) GOSF relates the approximate change in power ow in line i-j (i.e. between bus i and j) due to the outage of generator at bus k, whereas LOSF helps to calculate the approximate change in power ow in line i-j due to outage of line m-n. The generation outage sensitivity factor is dened by,
k ij =

fij Pk

(5.1)

k where, ij GOSF of line i-j for generation change at bus k

fij Change in power ow in line i-j Pk Change in generation at bus k

k The factor ij denotes the sensitivity of the line ow on line i-j due to change in generation at bus k. In equation (5.1), it is assumed that the generation lost at bus k would be exactly compensated by the reference or slack bus. Now, if the generation at bus k was generating an 0 0 amount of power equal to Pk , then to represent the outage condition, Pk = Pk .

Hence, the new power ow over the line i-j would be given as,
k 0 0 0 0 k n ij Pk Pk = fij = fij + fij = fij + ij fij

(5.2)

k The factor ij would be pre-calculated and stored in the memory. As we will see later, the values k of ij depend only on the network parameters and therefore, are constant. However, it should be k m noted that for any particular line i-j, the factors ij and ij (for generation outage at bus m) are dierent and therefore need to be pre-calculated separately. Once these factors are pre-calculated and stored, the new values of line ow over any line can easily be estimated very quickly from equation (5.2). If the new power ow over any line is found to be more than the corresponding limit, then the operator can be alerted for taking an appropriate pre-emptive action.

In equation (5.2), it is assumed that the lost generation at bus k would be taken up by the slack bus. However, it is also quite possible that the lost generation would be compensated by all the remaining on-line generators combinedly, in which, each of the on-line generators would take up some fraction of the lost generation in some particular ratio. One of the most frequently used methods assumes that the on-line generators share the lost generation in proportion to their maximum MW rating. Thus, the proportion of generation picked up by generation g is given by 227

g k, gk =

Pgmax
M a=1 k max Pa

(5.3)

Where, M Total number of generators in the system

gk Proportionality factor for generation g to pick up generation when unit k fails max Maximum MW rating for generator a. Pa
Now, as the sensitivity factors shown in equation (5.1) are linear in nature, the eects of simultaneous generation change in several generators on a particular line can be obtained by following superposition principle. Hence, the new line ow in the line i-j becomes,
(n) (0) k fij = fij + ij Pk m a=1 a ij Pa ak

(5.4)

In equation (5.4) it is assumed that no remaining on-line generation hits the generation limit. The line outage distribution factors are also dened similarly. The LOSF is dened by,

ij, mn =

fij
(0) fmn

(5.5)

Where, ij, mn Line outage distribution factor for line i-j under outage of line m-n.
(0) fmn Power ow over line m-n in the pre-outage condition.

Therefore, for the outage of line m-n, the new ow over line i-j is given by,
(n) (0) (0) fij = fij + ij, mn fmn

(5.6)

Again, as we will show later, the factors ij,mn are constant as they are dependent only on the line parameters. Therefore, they would be pre-calculated and stored in the memory. As a result, for the outage of any line m-n, the new power ows over all the other lines can be estimated very quickly.

5.2

DC load ow and generation outage distribution factor

We have already discussed the concepts of linear sensitivity factors. These factors are calculated based on the concept of DC power ow and hence, let us rst have a look at DC power ow technique.

5.2.1

DC power ow

From FDLF method, we know, P = [B ] , where each elements of the matrix [B ] is negative BUS bus matrix. Now, in DC power of the imaginary parts of the corresponding elements of the Y 228

ow analysis, apart from using the above decoupled relation between P and , several other simplifying assumptions are also taken as follows: a. The system is lossless (i.e. line resistance is neglected) and therefore, each line is represented by its reactance only. b. The voltage of each bus is maintained at the rated voltage, i.e. 1.0 p.u. c. For any line m-n, the angular dierence between its terminal buses is quite small, and hence, cos m cos n (as m n ) and sin(m n ) (m n ) rad (as m n 0) . With these assumptions, the power ow over a line becomes,

Pij =

Vi Vj 1 sin(i j ) (i j ) (p.u) xij xij

(5.7)

In equation (5.7), the quantity xij denotes the reactance of the line i-j. From this equation it is observed that the line power ow is basically a linear combination of the terminal bus voltage angles. Moreover, the current ow over the line i-j is given by,

ij = Vi Vj I jxij (Vi cos i Vj cos j ) + j (Vi sin i Vj sin j ) = jxij 1 1 = (sin i sin j ) (i j ) xij xij

(5.8)

Equation (5.8) has been written under an added assumption that both the angles i and j are individually quite small in magnitude. From equations (5.7) and (5.8) it is observed that in DC power ow model, the expressions of line power ow and line current are same in per unit. We will utilize this fact for computing LOSF in future. In the next lecture, we will discuss a method for calculating GOSF.

229