Upload
vunhi
View
216
Download
2
Embed Size (px)
Citation preview
DEALING WITH TWO TIME SCALES IN DISTRIBUTION SYSTEM STATE ESTIMATORS
Panel session State Es(ma(on for Distribu(on Opera(ons: sharing the experiences of implementa(on, usage and complexi(es
A. Gómez-‐Expósito C. Gómez-‐Quiles University of Seville
Spain
I. Dzafic Siemens AG Germany
IEEE/PES General Mee(ng, Vancouver July 25, 2013
CONTENTS
• Mo(va(on • Distribu(on-‐level informa(on • State Es(ma(on with two (me scales • Preliminary results • Conclusions
© A. Gómez-‐Expósito, 2013
Telemetered informa(on in today’s distribu(on substa(ons
Feeder head currents
Bus voltage magnitudes
MOTIVATION
© A. Gómez-‐Expósito, 2013
MOTIVATION
The consequence is...
Feeder head currents
Bus voltage magnitudes
MV & LV feeder system is literally a
BLACK BOX
© A. Gómez-‐Expósito, 2013
MOTIVATION
Is it sa(sfactory?
Passive loads Planning criteria are sufficient
© A. Gómez-‐Expósito, 2013
MOTIVATION
Is it sa(sfactory?
?
Ac(ve loads
?
?
?
Need to check (in real-‐(me): -‐ Overvoltages -‐ Undervoltages -‐ Feeder conges(ons -‐ Islanding, etc.
© A. Gómez-‐Expósito, 2013
MOTIVATION
What can we do to improve drama(cally the situa(on?
Use AMI/AMR data to “illuminate” the MV radial system
Smart grid paradigm
© A. Gómez-‐Expósito, 2013
DistribuKon-‐level informaKon
• SCADA & DMS • Feeder automa(on devices • Distributed generators • Historic load pagerns/profiles • AMI data (Smart Meters concentrators)
Several (heterogeneous) informa(on sources in upcoming Smart Grids:
© A. Gómez-‐Expósito, 2013
DistribuKon-‐level informaKon
• SCADA & DMS • Feeder automa(on devices • Distributed generators • Historic load pagerns/profiles • AMI data (Smart Meters concentrators)
• Very few RTU measurements captured at HV-‐MV substa(ons
• Sampled every few seconds
© A. Gómez-‐Expósito, 2013
DistribuKon-‐level informaKon
• SCADA & DMS • Feeder automaKon devices • Distributed generators • Historic load pagerns/profiles • AMI data (Smart Meters concentrators)
• Intermediate switching points for fault management
• Can be used as addi(onal telemeasured points (RTUs)
© A. Gómez-‐Expósito, 2013
Example of feeder automaKon devices Siemens Feeder Automa(on controller: SIPROTEC 7SC80
Can be used both as fault locator and RTU (V, I, P, Q, cos ϕ)
DistribuKon-‐level informaKon
• SCADA & DMS • Feeder automa(on devices • Distributed generators • Historic load pagerns/profiles • AMI data (Smart Meters concentrators) Depending on specific regula(on:
• Day-‐ahead hourly forecas(ng of energy produc(on
• Real produc(on periodically submiged to DMS
© A. Gómez-‐Expósito, 2013
DistribuKon-‐level informaKon
• SCADA & DMS • Distribu(on automa(on devices • Distributed generators • Historic load paUerns/profiles • AMI data (Smart Meters concentrators) • Day-‐ahead hourly load forecas(ng
• Assumed PF for typical loads -‐ Hourly values of P&Q
© A. Gómez-‐Expósito, 2013
DistribuKon-‐level informaKon
• SCADA & DMS • Distribu(on automa(on devices • Distributed generators • Historic load pagerns/profiles • AMI data (Smart Meters concentrators)
• Involves DMS-‐AMI communica(ons • Depending on bandwidth availability: -‐ From 15’ to 24h snapshot latency
© A. Gómez-‐Expósito, 2013
Two different latencies
• SCADA & DMS • Feeder automa(on devices • Distributed generators • Historic load pagerns/profiles • AMI data (Smart Meters concentrators)
• Snapshots updated from few sec. to about a minute
• Insufficient to assure network observability
© A. Gómez-‐Expósito, 2013
Two different latencies
• SCADA • Feeder automa(on devices • Distributed generators • Historic load pagerns/profiles • AMI data (Smart Meters concentrators)
• Snapshots updated from 15’ to 24h • Barely cri(cal informa(on to assure
network observability © A. Gómez-‐Expósito, 2013
Two different latencies
• SCADA • Feeder automa(on devices • Distributed generators • Historic load pagerns/profiles • AMI data (Smart Meters concentrators)
Minimum redundancy levels obtained only when both informa(on types are properly combined
© A. Gómez-‐Expósito, 2013
State EsKmaKon with two measurement latencies
Two dis(nct sets of measurements, captured at different rates:
t
zp,j zp,j+1
zr,k
Real-time incomplete information system (fast rate)
Pseudomeasurements: less accurate, “complete” information system (slow rate)
zr,k+1
!T = nTk
kT
© A. Gómez-‐Expósito, 2013
zp,j
zp,j+1
zr,k
Accuracy of Zp,j depends on the rate of change of loads
zr,k+1
Load evolu(on
More accurate Less accurate
State EsKmaKon with two measurement latencies
© A. Gómez-‐Expósito, 2013
Accuracy of Zp,j depends on the rate of change of loads
State EsKmaKon with two measurement latencies
zp,j
zp,j+1
Load evolu(on
λ: Rate of load change between two consecu(ve slow snapshots
© A. Gómez-‐Expósito, 2013
MathemaKcal model: WLS es(ma(on with zp and zr
⎥⎦
⎤⎢⎣
⎡+⎥⎦
⎤⎢⎣
⎡=⎥
⎦
⎤⎢⎣
⎡
r
p
r
p
r
p
xhxh
zz
ε
ε
)()(
State EsKmaKon with two measurement latencies
Gauss-‐Newton methodology (Normal equa(ons): where
( ) ( ) ( ))()( xhzWHxhzWHxHWHHWH rrrtrppp
tprr
trpp
tp −+−=Δ+
cov(!p ) =Wp
!1 cov(!r ) =Wr
!1
© A. Gómez-‐Expósito, 2013
• First (cold) execuKon: both zp and zr updated
State EsKmaKon with two measurement latencies
t
zp,j zp,j+1
zr,j
Base-‐case load: P & Q
© A. Gómez-‐Expósito, 2013
• n-‐1 “warm” execuKons: only zr updated
State EsKmaKon with two measurement latencies
t
zr,k
zp,j zp,j+1
Base-‐case load: P & Q
Latest es(mate used to start itera(ons © A. Gómez-‐Expósito, 2013
• n-‐1 “warm” execuKons: only zr updated
State EsKmaKon with two measurement latencies
t
zr,k
zp,j zp,j+1
Base-‐case load: P & Q
Final load: λ(P & Q)
Latest es(mate used to start itera(ons © A. Gómez-‐Expósito, 2013
Case study
Can a small subset of “real-‐(me” measurements,
when combined with a cri(cal set of (outdated)
pseudomeasurements, provide accurate enough
es(mates ? (reconstruc(on of network state)
© A. Gómez-‐Expósito, 2013
Case study
• Small radial distribu(on system (20 buses)
• Pseudomeasurements (zp): V1 and P & Q at all buses
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Nx= Nzp = 39
© A. Gómez-‐Expósito, 2013
Case study
• Small radial distribu(on system (20 buses)
• Pseudomeasurements (zp): V1 and P & Q at all buses
• Real-‐Kme measurements (zr): I & V 2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Current flow at branch 1-‐2
Voltage mag. at node 5
Nx= Nzp = 39
Nzr = 5
Redundancy = 1.128
• Very few in prac(ce • In our tests limited to:
© A. Gómez-‐Expósito, 2013
Case study
• Small radial distribu(on system (20 buses)
• Pseudomeasurements (zp): V1 and P & Q at all buses
• Different scenarios for real-‐Kme measurements (zr): • Measurement type: I & V • Measurement loca(ons • Rate of load change: λmax=1.5 • Change in power flow direc(on
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Current flow at branch 1-‐2
Voltage mag. at node 5
© A. Gómez-‐Expósito, 2013
Case study
• Small radial distribu(on system (20 buses)
• Pseudomeasurements (zp): V1 and P & Q at all buses
• Different scenarios for real-‐Kme measurements (zr): • Measurement type: I & V • Measurement loca(ons • Rate of load change: λmax=1.5 • Change in power flow direc(on
• WLS solu(ons compared with exact state: • zp values “frozen” with base-‐case load • zr values updated according to load evolu(on
[1< λ<1.5 ; ∆λ=0.05]
© A. Gómez-‐Expósito, 2013
-‐ A -‐ zr = currents
-‐ B -‐ zr = voltages
Test 1: Comparison of measurement type: I versus V
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
Set zr: uniformly distributed sets of measurements
-‐ Current meas. beger than voltage meas. (up to 10 (mes lower voltage error) -‐ Errors increase linearly with load growth (pseudomeasurement obsolescence)
0,0E+00
2,0E-‐03
4,0E-‐03
6,0E-‐03
8,0E-‐03
1,0E-‐02
1,2E-‐02
1,1 1,2 1,3 1,4 1,5
A) Average errors
B) Average errors
A) Maximum errors
B) Maximum errors
λ
|Viwls !Vi
exact |
© A. Gómez-‐Expósito, 2013
-‐ A -‐ zr uniform
-‐ C -‐ zr at the ends
Test 2: Comparison of measurement loca(on
Sets zr: current measurements, different loca(ons
-‐ Beger results with uniformly distributed current measurements
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
λ
0,0E+00
2,0E-‐04
4,0E-‐04
6,0E-‐04
8,0E-‐04
1,0E-‐03
1,2E-‐03
1,4E-‐03
1,6E-‐03
1,1 1,2 1,3 1,4 1,5
A) Average errors
C) Average errors
A) Maximum errors
C) Maximum errors
|Viwls !Vi
exact |Voltage errors
© A. Gómez-‐Expósito, 2013
-‐ A -‐ zr uniform
-‐ C -‐ zr at the ends
Test 2: Comparison of measurement loca(on
-‐ Beger results with uniformly distributed current measurements
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
λ
0,0E+00
1,0E-‐01
2,0E-‐01
3,0E-‐01
4,0E-‐01
5,0E-‐01
6,0E-‐01
7,0E-‐01
8,0E-‐01
1,1 1,2 1,3 1,4 1,5
A) Average errors
C) Average errors
A) Maximum errors
C) Maximum errors
Sets zr: current measurements, different loca(ons
| Siwls ! Si
exact |Injec(on errors
© A. Gómez-‐Expósito, 2013
Test 2: Comparison of measurement loca(on
1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 3-17 5-18 18-19 19-20 20-21 21-22 19-23 10-780
2
4
6
8
10
12
14
16
18
20
Sij (p
er u
nit,
Sba
se=1
00kV
A)
Branch i-j
λ = 1.1
0
0.05
0.1
0.15
0.2
0.25
Rel
ativ
e er
ror (
per u
nit)
Relative errorLoad FlowWLS SE
1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 3-17 5-18 18-19 19-20 20-21 21-22 19-23 10-780
2
4
6
8
10
12
14
16
18
20
Sij (p
er u
nit,
Sba
se=1
00kV
A)
Branch i-j
0
0.05
0.1
0.15
0.2
0.25
Rel
ativ
e er
ror (
per u
nit)
Relative errorLoad FlowWLS SE
-‐ A -‐ zr uniform
-‐ C -‐ zr at the ends
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Branch power flows
© A. Gómez-‐Expósito, 2013
Test 2: Comparison of measurement loca(on
1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 3-17 5-18 18-19 19-20 20-21 21-22 19-23 10-780
2
4
6
8
10
12
14
16
18
20
Sij (p
er u
nit,
Sba
se=1
00kV
A)
Branch i-j
λ = 1.5
0
0.05
0.1
0.15
0.2
0.25
Rel
ativ
e er
ror (
per u
nit)
Relative errorLoad FlowWLS SE
1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 3-17 5-18 18-19 19-20 20-21 21-22 19-23 10-780
2
4
6
8
10
12
14
16
18
20
Sij (p
er u
nit,
Sba
se=1
00kV
A)
Branch i-j
0
0.05
0.1
0.15
0.2
0.25
Rel
ativ
e er
ror (
per u
nit)
Relative errorLoad FlowWLS SE
-‐ A -‐ zr uniform
-‐ C -‐ zr at the ends
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Branch power flows
© A. Gómez-‐Expósito, 2013
-‐ A -‐
Test 3: Weigh(ng coefficient influence
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
Set zr: current measurements, uniformly distributed
0,0E+00
2,0E-‐04
4,0E-‐04
6,0E-‐04
8,0E-‐04
1,0E-‐03
1,2E-‐03
1,1 1,2 1,3 1,4 1,5
A) Average errors
D) Average errors
A) Maximum errors
D) Maximum errors
λ
σp = 0.1
σr = 0.1 σp = 0.1
σr = 0.001
-‐ Similar results with equal or different weights
-‐ D -‐
|Viwls !Vi
exact |Voltage errors
© A. Gómez-‐Expósito, 2013
Test 4: Influence of head current
-‐ A -‐ With I12
-‐ E -‐ Without I12
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
Same redundancy
-‐ Beger results with head current measurement
λ
0,0E+00
2,0E-‐04
4,0E-‐04
6,0E-‐04
8,0E-‐04
1,0E-‐03
1,2E-‐03
1,4E-‐03
1,1 1,2 1,3 1,4 1,5
A) Average errors
E) Average errors
A) Maximum errors
E) Maximum errors
|Viwls !Vi
exact |Voltage errors
© A. Gómez-‐Expósito, 2013
Test 5: Change in power flow direc(on (distributed genera(on)
Scenario: Power injec(on changes from P12 to -‐P12 while rest of loads increase from λ=1 up to λ=1.5.
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Set zr: currents at the end of feeder laterals
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.05
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.1
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.15
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.2
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.25
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.3
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.35
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.4
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.45
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.5
LF SWLS S
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
-‐ Inability of current measurements to track sign changes in P
© A. Gómez-‐Expósito, 2013
-‐ Same happens when Q12 changes its sign
-‐ C’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
Set zr: Ac(ve power flow measurements ‘ ’ instead of currents
Test 5: Change in power flow direc(on (distributed genera(on)
Scenario: Power injec(on changes from P12 to -‐P12 while rest of loads increase from λ=1 up to λ=1.5.
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1
LF SWLS S
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.05
LF SWLS S
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.1
LF SWLS S
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.15
LF SWLS S
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.2
LF SWLS S
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.25
LF SWLS S
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.3
LF SWLS S
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.35
LF SWLS S
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.4
LF SWLS S
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.45
LF SWLS S
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
-‐ F’ -‐
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21 22 23 78
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
P (p
er u
nit,
Sbas
e=10
0kVA
)
Node number
λ = 1.5
LF SWLS S
-‐ Ac(ve power flow measurements detect power flow inversion
Test 5: Change in power flow direc(on (distributed genera(on)
© A. Gómez-‐Expósito, 2013
Conclusions
• Smart grid context: new “real-‐(me” feeder measurements will become gradually available (fault detec(on & isola(on).
• Insufficient to achieve observability: need to combine with other informaKon sources (pseudomeasurements): AMI.
• Two measurement latencies and accuracy levels.
• Very few Ampere measurements may suffice to provide reasonable es(mates, even for 50% load increase.
• Inability of Ampere measurements to track counterflows.
• Power rather than Ampere measurements encouraged in the presence of distributed generaKon.
• Future efforts: test larger realis(c systems & prac(cal implementa(on (computa(onal saving).
© A. Gómez-‐Expósito, 2013
DEALING WITH TWO TIME SCALES IN DISTRIBUTION SYSTEM STATE ESTIMATORS
Panel session State Es(ma(on for Distribu(on Opera(ons: sharing the experiences of implementa(on, usage and complexi(es
A. Gómez-‐Expósito C. Gómez-‐Quiles University of Seville
Spain
I. Dzafic Siemens AG Germany
IEEE/PES General Mee(ng, Vancouver July 25, 2013
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
-‐A-‐ zr current meas.
-‐G-‐ zr ac(ve power meas.
Test 7: current measurements versus ac(ve power measurements
© A. Gómez-‐Expósito, 2013
-‐G-‐ zr ac(ve power meas.
1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 3-17 5-18 18-19 19-20 20-21 21-22 19-23 10-780
2
4
6
8
10
12
14
16
18
20
S ij (per
uni
t, Sb
ase=
100k
VA)
Branch i-j
λ = 1.5
0
0.05
0.1
0.15
0.2
0.25
Rela
tive
erro
r (pe
r uni
t)
Rel.errorLF SWLS S
1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 3-17 5-18 18-19 19-20 20-21 21-22 19-23 10-780
2
4
6
8
10
12
14
16
18
20
P ij (per
uni
t, Sb
ase=
100k
VA)
Branch i-j
λ = 1.5
0
0.05
0.1
0.15
0.2
0.25
Rela
tive
erro
r (pe
r uni
t)
Rel.errorLF SWLS S
2 3 4 5
1
6 7 8 9 10 11 12
17
18 19 20 21 22
23
78
© A. Gómez-‐Expósito, 2013
-‐A-‐ zr current meas.
1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 3-17 5-18 18-19 19-20 20-21 21-22 19-23 10-780
2
4
6
8
10
12
14
16
18
20
P ij (per
uni
t, Sb
ase=
100k
VA)
Branch i-j
λ = 1.5
0
0.05
0.1
0.15
0.2
0.25
Rela
tive
erro
r (pe
r uni
t)
Rel.errorLF SWLS S
1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 3-17 5-18 18-19 19-20 20-21 21-22 19-23 10-780
2
4
6
8
10
12
14
16
18
20
S ij (per
uni
t, Sb
ase=
100k
VA)
Branch i-j
λ = 1.5
0
0.05
0.1
0.15
0.2
0.25
Rela
tive
erro
r (pe
r uni
t)
Rel.errorLF SWLS S
2 3 4 5
1
6 7 8 9
10 11 12
17
18 19 20 21 22
23
78
© A. Gómez-‐Expósito, 2013