Transcript Methodology for Flexible, Cost-Effective Monitoring of Voltage Sags

Frankfurt (Germany), 6-9 June 2011
METHODOLOGY FOR FLEXIBLE,
COST-EFFECTIVE MONITORING OF
VOLTAGE SAGS
Manuel Avendaño
J. V. Milanović
School of
Electrical &
Electronic
Engineering
Manchester, UK
Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
What did we do?

Proposed methodology for determining a range
of best monitoring programmes for estimating
the performance of sags with different
characteristics.

Incorporated
user-defined
voltage
sag
characteristics and a measure of the overall
accuracy of sag estimation.
Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
Presentation Outline
Why did we do it? (Importance and motivation)
 How did we do it? (Methodology)
 What did we get? (Results)
 What did we learn? (Conclusions)

Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
Why did we do it?

Knowledge of voltage sag incidence in the
network can help in tailoring solutions to mitigate
the consequences of sags.

Estimation of sag characteristics is required
when measurements are not available.

Fault location method utilized directly influences
the number of monitors.
Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
Why did we do it?

Sag monitoring programs (SMPs) should be
focused on quantifying most critical sags
−

(E.g. SARFI-90%, SARFI-70%, SEMI F47, etc)
To provide a measure for assessing the sag
estimation derived from a SMP
−
(Diff. between real and estimated events)
Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
How did we do it?

Selection of monitor locations based
minimization of overall sag estimation error.

Utilization of existing fault location method.

Application in a generic distribution system
(GDS) and comparison with an optimal
placement method.
Manuel Avendaño – UK – Session 2 – Paper 0529
on
Frankfurt (Germany), 6-9 June 2011
Sag estimation error (SEE)
N
SEE( SEMI F 47) 
N

i 1
i , real
i , estimated
X SEMI
X
F 47
SEMI F 47
N
= total number of buses
i , real
= real number of sags below i.c. SEMI F47 at bus i
X SEMI
F 47
i , estimated
X SEMI
F 47 = estimated number of sags below i.c. SEMI F47 at bus i
SEMI F47 can be substituted by any other voltagetolerance curve (CBEMA), performance index (SARFI), etc.
Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
Monitor placement
1.
2.
3.
4.
5.
6.
Set target value for SEE or number of monitors (stop
criteria).
Simulate faults to obtain sag performance.
Perform fault location using voltage measurements of
all buses.
Calculate SEE incurred by all buses.
Monitor location = min(SEE)
Repeat steps 3-5 until a stop criterion is fulfilled.
Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
What did we get?
An iterative search algorithm that is:

Flexible. One or multiple monitoring programmes can
be determined for any kind of user-defined voltage sag
characteristics.

Cost-effective.
If
technical
and/or
economic
constraints limit the number of monitors to be deployed,
a series of SMPs can be provided accordingly.
Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
Application

295-bus GDS, 278 lines, 37 transformers.
Voltage level
(kV)
Fault clearing
time (cycles)
11
18
33
9
132
Bus faults
4.8
3.6
Percent of nominal voltage
100
80
60
40
20
0
GDS
sag characteristics
5
10
15
20
25
30
Duration (cycles)
35
40
45
50
GDS equipment shut-down region below SEMI F47
Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
Sag Monitoring Programmes
SEE (number of sags)
Number of
monitors
SARFI-90
SARFI-80
SARFI-70
SEMI F47
1
876
873
876
541
2
459
459
459
218
3
255
255
313
117
4
118
118
119
62
5
22
23
24
16
6
11
12
13
5
7
5
10
12
3
8
3
8
8
1
9
1
6
6
0
10
0
0
0
0
Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
Reduction of sag estimation error
Sag Estimation Error
1200
1000
800
600
400
200
0
1
2
3
4
1
2
3
4
30
25
20
15
10
5
0
5
5
6
6
7
8
7
8
Number of monitors
Manuel Avendaño – UK – Session 2 – Paper 0529
9
9
10
10
Frankfurt (Germany), 6-9 June 2011
Location of monitors
SMP – SARFI-90
Manuel Avendaño – UK – Session 2 – Paper 0529
Optimal monitoring
Frankfurt (Germany), 6-9 June 2011
Voltage sag magnitude (p.u.)
Effects of robustness in fault location
method on sag magnitude estimation
1
0.9
Est. 5 mon
Real
Est. 12 mon
0.8
0.7
0
50
100
150
Buses
Manuel Avendaño – UK – Session 2 – Paper 0529
200
250
300
Frankfurt (Germany), 6-9 June 2011
Comparison with optimal monitoring
12 monitors optimally placed vs. 5 monitors placed
with proposed approach.
Real
Est. 5 mon
Est. 12 mon
SARFI-90 index
15
10
5
0
0
50
100
150
Buses
Manuel Avendaño – UK – Session 2 – Paper 0529
200
250
300
Frankfurt (Germany), 6-9 June 2011
Comparison with optimal monitoring
Distribution of SEE for Monte Carlo simulations
representing 100 years of system performance
Sag Estimation Error
30
25
20
15
10
5
0
1
2
SMP (5 monitors)
OSMP (12 monitors)
Manuel Avendaño – UK – Session 2 – Paper 0529
Frankfurt (Germany), 6-9 June 2011
Conclusions
A methodology for determining a range of best
voltage sag monitoring programmes is
proposed.
 DNOs can choose a sag monitoring programme
specifically
designed
to
estimate
the
performance of the sags more relevant to its
customers.
 Due to the fault location technique employed it is
more robust than previous approaches.

Manuel Avendaño – UK – Session 2 – Paper 0529