Professional Documents
Culture Documents
NC
compared with the Maximum MWM of the system without N
any contingency (MMWM). To rank contingencies by their M= l (2)
impact on voltage collapse, we arbitrarily propose a boundary l =0
of 90% to 100% of MMWM for Acceptable contingencies.
By ranking the contingencies, we are able to The zero level contingency, NC0 , means no element in the
substantially reduce the number of contingencies out of all system is subject to failure. To simplify writing of Equation
possible contingencies that need to be considered for voltage 2, we define NC0,N to represent the sum of all contingencies
stability analysis. Three categories are proposed for all from zero level to N level, NC0 , NC1 , NC2 , , NCN as in
possible contingencies based on the MWM reduction from Equation 5.
MMWM. The three proposed categories are:
M = NC0, N (3)
The number of lth level contingencies is defined as:
( N )!
NCl = (4) Base Case Data
( l )!*( N l )!
2 [ Load ] =[ Load ]+
[ Delta Load ]
NC0,2 = NCl =1 + N + ( 1 )( N )( N 1) (5)
l =0
2
Yes
The proposed algorithm presented in this paper saves Does the Load-Flow
enormous computing time whenever many load flow converge?
simulations are needed in voltage collapse studies. The
No
number of iterations can be reduced significantly using the No
proposed algorithm outlined in flowchart shown in Figure 2. [ Load ] = [ Load ] - 2 [ Delta Load ]
Using the new algorithm in Figure 2, most of the computation
[ Delta Load ] = 1/2 [ Delta Load] Yes
time is spent on measuring the exact value of MWM for
Significant and Unacceptable contingencies, and less time is Does the
Load-Flow
spent on measuring the boundary of MWM for Acceptable Yes
converge?
contingency. The arbitrary boundary of 90% to 100% of
MMWM is found for Acceptable contingencies in one step Is Max [ Delta Bus ] >
and because most of the contingencies for large-scale systems 1 MW ?
do normally belong to the Acceptable category, the algorithm No Perform Load-Flow
is fast and practical in determining those contingencies that
need to be included in the study. The amount of time that is MWM i = SUM [ Load ] SUM [Load Base]
saved to find the boundary of 90% to 100% for Acceptable (Ci) is
MWM1 = MMWM Acceptable
contingencies depends on the reliability of the system for Contingency
more reliable systems there are more Acceptable
contingencies and therefore more computing time is saved to
categorized the contingencies. The 90% value is chosen as the SORT [ MWM ]
New Load = 90% MMWM
threshold in our study, although it may change from one study
to another.
The algorithm starts by finding the MMWM for the Is
base case (without any contingency) and the user sets an MWM Negative?
arbitrary Acceptable MWM threshold, for example 90% of the Next Contingency (Ci)
No
MWMM. If the simulated contingencys MWM is within the Yes
specified threshold, the contingency is ranked as Acceptable,
otherwise it is categorized as either Unacceptable or (Ci) is Unaccptable
Significant. For the Acceptable contingencies, there is no Contingency
need to find the exact MWM since it is greater than 90% of
the MMWM. All Acceptable contingencies can be treated as
(Ci) is Significant
a hypothetical base case without any contingency since their Contingency
MWM is more than 90% and not far from MMWM.
Following the filtering out of the Acceptable Figure 2: the Flowchart for ranking contingencies
contingencies, the remaining contingencies are tested against
the base case. If the simulated contingency has a solution for The new algorithm starts with the simulation of
the base case, the contingency is ranked as Significant, if not, voltage collapse at base load in system and without any
it is categorized as Unacceptable contingency. Using the fast contingency and finds the MMWM by performing several
ranking method of the proposed algorithm, within two load flow iterations. The MMWM is defined as:
iterative steps the Unacceptable contingencies are identified
and later measured for their MWM. MMWM = Pmax Pbase (6)
Then the range of 90% to 100% of MMWM is set for The voltage collapse occurs in Area 2 at a total load
Acceptable contingencies, 0% to 90% of MMWM for of 7443 MW ( Pmax ). The MMWM for the system is:
Significant contingencies, and finally the contingencies with
negative margin are called Unacceptable contingencies in this MMWM =7443 5677 = 1766 MW
paper. The next run of the algorithm calls the first case with
The MMWM of 1766 MW is available to serve 1766
contingency and the simulation starts not from the base load,
MW of load increase if all the generators, transmission lines,
but from the load point at the boundary of the Significant and
transformers, and all other components are in service and
Acceptable contingencies. This load point is at the threshold
operate properly.
of the Acceptable contingency with 90% of the MWMM
greater than the base load. In the first iteration of the B. Contingency Ranking of the Test System
simulation to obtain MWM, if the load flow solution
There are 177 transmission lines in the 118-bus test
converges, the contingency falls in the Acceptable category
system and, considering the outage of each single line at a
and the next simulation starts for the next contingency. This
time, there are 177 single contingency cases.
feature saves the time by placing the Acceptable contingency
out of the list for the remaining contingencies, and then the Table 1: MW Margin
next contingency is simulated. If the load flow solution
doesnt converge, the complete set of load flow solution is Contingency MW
Pmax
Case #
performed to find the exact value of MWM for that Line Outage MW Margin
(MW)
contingency. The first portion of the load flow finds the nose From Bus To Bus Margin (%)
point of the P-V curve for each contingency by increasing the
1 9 10 4730 -947 -54%
load stepwise upward to a point where no extra load can be
supplied by the system. If the load flow converges then the 2 8 9 4730 -947 -54%
system load (Load) is increased by a small load increment 3 26 30 6706 1029 58%
(Delta Load) of 1% of the base load (Base Load). Otherwise, 4 4 5 6901 1224 69%
the system load is increased by smaller increments of 0.5% of 5 100 103 7033 1356 77%
Delta Load. The steps toward the nose of the P-V curve get 6 25 27 7145 1468 83%
smaller as the solution gets closer to the nose. To avoid
7 23 25 7165 1488 84%
unnecessary running of the load flow program for very little
adjustment, the load increment will not get smaller than 1 8 69 75 7259 1582 90%
MW. When the load increment becomes less than or equal to 9 38 65 7268 1591 90%
1 MW, and the load flow program does not converge to a 10 22 23 7297 1620 92%
solution, the nose point of the P-V curve for that particular 11 17 113 7310 1633 92%
case is calculated. The load at the nose point of the P-V curve 12 3 5 7341 1664 94%
is considered to be in the middle of the maximum load point
13 11 13 7341 1664 94%
with a load flow solution and the next load point without a
load flow solution. The MWM of the case is the difference of 14 8 30 7344 1667 94%
the total load of all the buses and the base load. In the last 15 88 89 7346 1669 95%
step, the MWM for each contingency case is put in a sorted 16 17 18 7347 1670 95%
array to be categorized. 17 64 65 7351 1674 95%
IV. SIMULATION RESULTS 18 15 17 7348 1671 95%
19 30 38 7360 1683 95%
A. IEEE 118-bus Test System
20 21 22 7361 1684 95%
The simulation was initially run by conventional P-V 21 4 11 7363 1686 95%
curve calculation to find the load margin for all possible first
22 23 32 7364 1687 96%
level contingencies, and then was tested using the proposed
screening method. The program was successfully able to 23 5 6 7381 1704 96%
categorize the same contingencies in comparatively shorter 24 1 3 7384 1707 97%
time with no error. 25 2 12 7388 1711 97%
The complete description of IEEE 118-Bus Test
System can be found in reference [20]. In the base-case, the 178 No Contingency 7443 1766 100%
total system loading is 5677 MW, the swing bus Sporn (bus
Using the flowchart in Figure 2, all of the 177 single
# 69) generates 516 MW, which is the largest unit following
contingency cases are simulated. In Figure 2, the first run is
ClinchRv (bus # 89) with 607 MW at generation. The
the simulation of the system without any contingency. The
minimum voltage is at bus Darrah (bus # 76) with 0.943pu.
second run is the simulation of the system with the
transmission line from buses 1 (Riverside) to bus 2 (Pokagon)
out of service. The program repeats testing all other 176 cases
likewise. At the end of the simulation, the MW Margin for all Table 2: Ranking
of the 178 cases the base case and 177 single contingency
cases - are sorted in ascending order and saved in a vector, C o ntingency Lo ad
Case #
called [MWM] in the flowchart. After sorting the MWMs, the L ine O utage M argin R anking
base case with the maximum MWM of 1766 is saved at the Fro m B us T o B u s D ecrease
end of the list (C178 ). The first two cases in the list, C1 and 1 9 10 1 54%
U na ccepta ble
C2, have a negative MWM. Table 1 summarizes the result of 2 8 9 1 54%
the simulation. The table is truncated for the cases with more 3 26 30 42 %
than 1711 MW Margin. The first 25 cases in Table 1 4 4 5 31 %
correspond to the removal of one transmission line at a time as 5 1 00 103 23 %
described in the second and the third column of the table. The 6 25 27 17 % Significant
Maximum MW load served (before a loss of voltage due to 7 23 25 16 %
each contingency occurs) is shown in the fourth column. 8 69 75 10 %
Columns five and six indicate the MW and the percent MW
9 38 65 10 %
Margin respectively. According to Table 1, the removal of line
10 22 23 8%
9-10 or line 8-9 will have the most severe impact on loss of
11 17 113 8%
the system voltage. While the removal of line 2-12 (Case 25)
will hardly affect the voltage since 97% of the MMWM is still 12 3 5 6%
available even without line 2-12 being in service. 13 11 13 6%
The ranking of first level contingencies based on the 14 8 30 6%
impact of the contingency on the MW Margin and moving 15 88 89 5%
closer to the point of voltage collapse is shown in Table 2. 16 17 18 5%
Column 4 in Table 2 shows the load margin percentage 17 64 65 5%
decrease from the MWM of the system. The negative MW 18 15 17 5%
Margin for the Unacceptable first level contingencies is such A ccepta ble
19 30 38 5%
that, when line 9-10 or line 8-9 is out of service, the supply is 20 21 22 5%
less than the demand. In this case, the load flow solution does
21 4 11 5%
not converge and a loss of voltage at some buses in the system
is expected. Table 2 demonstrates that IEEE 118-Bus Test 22 23 32 4%
System has 168 first-level contingencies with less than 10% 23 5 6 4%
MW Margin reduction from the MW Margin of the case 24 1 3 3%
without any contingency. That means 95% 25 2 12 3%
[(168/178)*100=95%] of the total possible first level ...
contingencies have less than 10% impact on the systems 178 N o C o nting ency 0%
MWM, so they are called Acceptable in this paper. The rest
of possible contingencies (5%) has higher impact and falls in a
short list for more detailed studies. The simulation shows that
for IEEE 118-Bus Test System, 1% of the first level C. Large scale Test System
contingencies are Unacceptable, and 4% are Significant.
The system under investigation is a real size power
Figure 3 shows the percentage of Unacceptable, Significant,
system with more than 25,000 buses, and is divided into many
and acceptable contingencies using a pie chart.
areas based on the geographical parameters. This paper is
focused only on areas 1, 2 and 3 with 505 buses, and 589
transmission lines and generators in service.
1%
Figure 3: Pie Chart for three contingency categories Using Equation 5 for 589 elements in the system,
there are 173,755 possible first and second level
contingencies. The load flow program was applied to simulate As it was mentioned before the time needed to run the
the voltage collapse in the system and 173,755 contingencies program and also the size of the output report are so large that
at different load levels were considered. The results for the makes the analysis almost impractical.
contingencies with MWM less than MMWM are shown in
Table 4. Using the flow chart in Figure 2, the simulation took
only 4 hours and the output is very short and easy to analyze. Table 4 (continued): MWM for Large scale Power System
The total load for the three areas can be increased from 2911
MW to 4195 MW before occurrence of any voltage collapse
Contingency Level
Line Outage
assuming there is no failure in the system. The system will
Pmax (MW)
Base Load
MWM(j)
collapse when the load reaches to 4197 MW and at this point
the MMWM, from Equation 6, is 1286.
From Bus
j
To Bus
MMWM = 4197 2911 = 1286
Table 4: MWM for Large scale Power System
451 461
590 2 2805 2911 -106
Contingency Level
Base Load
451 461
MWM(j)
591 2 2805 2911 -106
453 457
From Bus
j
To Bus
451 461
592 2 2805 2911 -106
453 522
451 461
593 2 2805 2911 -106
1 0 No Contingency 4200 2911 1289 453 530
1 1 451 461 3080 2911 169 451 461
594 2 2805 2911 -106
2 1 452 461 3080 2911 169 453 532
452 461
3 1 461 468 3080 2911 169 595 2 3080 2911 169
453 469
4 1 461 468 3080 2911 169
451 461
5 1 461 538 3080 2911 169 596 2 3080 2911 169
454 469
6 1 463 468 3100 2911 189 451 461
7 1 476 478 3100 2911 189 597 2 3080 2911 169
454 514
8 1 481 482 3100 2911 189 451 461
598 2 3080 2911 169
9 1 481 512 3100 2911 189 455 463
10 1 500 503 3100 2911 189 451 461
599 2 3080 2911 169
11 1 500 524 3100 2911 189 455 468
12 1 502 524 3100 2911 189 451 461
600 2 3080 2911 169
13 1 504 511 3100 2911 189 456 460
451 461
14 1 505 506 3100 2911 189 601 2 3080 2911 169
466 520
15 1 505 523 3100 2911 189
451 461
16 1 511 525 3100 2911 189 602 2 3080 2911 169
467 531
17 1 516 517 3100 2911 189 451 461
18 1 516 525 3100 2911 189 603 2 3080 2911 169
467 533
19 1 516 517 3100 2911 189 451 461
604 2 3080 2911 169
20 1 516 525 3100 2911 189 469 470
21 1 517 523 3120 2911 209 ... ... ... ... ... ... ...
22 1 530 536 3120 2911 209 836 716
45791 2 3750 2911 839
23 1 536 537 3120 2911 209 838 716
24 1 574 716 3120 2911 209
589 1 16584 16685 4190 2911 1279
The threshold for Acceptable margins is assumed to be 90%
of the MMWM. Table 5 shows the threshold for Acceptable
contingencies for the three areas.
Table 5: Point of Collapse and Acceptable Threshold category holds more than 95% of the total possible
contingencies. Using the Acceptable, Significant, and
Area Pbase Pmax MMWM 90% of Acceptable Unacceptable categories helped us developing the algorithm
# that reduces the computation time by about 90%. The time
(MW) (MW) MMWM Threshold
saved by using the developed algorithm is very necessary for
(MW) large-scale power system analysis with thousands of elements.
The output of the simulation program is relatively shorter and
1 1145 1651 506 455 1600 the results are easier to analysis. The proposed ranking
scheme is valid for real power systems just by adjusting the
2 1251 1803 552 497 1748 Acceptable margins threshold.
The Authors
Majid Poshtan (1963) received his BSE degree from Tehran
University, Tehran/Iran, MSE degree from University of New
Brunswick, NB/Canada, and Ph.D. degree from Tulane
University, LA/USA in 2000. He taught power systems at
Tulane University, LA/USA until 2003 when he joined
Petroleum Institute AD/UAE as assistant professor. He has
been working in the area of large-scale power systems,
probabilistic reliability analysis, and Electric machines since
1996. He is a member of IEEE. (Petroleum Institute, PO Box
2533, AD, UAE, mposhtan@pi.ac.ae).