Exploring Complex Energy Networks
Florian Dorfler
@ETH for “Complex Systems Control”
compute
actuatethrottle
sensespeed
1 / 22
@ETH for “Complex Systems Control”
system
control
“Simple” control systems are well understood.
“Complexity” can enter in many ways . . .
1 / 22
A “complex” distributed decision making system
. . .
physical interaction
local subsystems and control
sensing & comm.
2
10
30 25
8
37
29
9
38
23
7
36
22
6
35
19
4
3320
5
34
10
3
32
6
2
31
1
8
7
5
4
3
18
17
26
27
28
24
21
16
1514
13
12
11
1
39
9
local system
local control
local system
local control
Such distributed systems include large-scale physical systems, engineeredmulti-agent systems, & their interconnection in cyber-physical systems.
2 / 22
Timely applications of distributed systems controloften the centralized perspective is simply not appropriate
Engineered multi-agent systems
Embedded robotic systems and sensor networks for
high-stress, rapid deployment — e.g., disaster recovery networks
distributed environmental monitoring — e.g., portable chemicaland biological sensor arrays detecting toxic pollutants
autonomous sampling for biological applications — e.g.,monitoring of species in risk, validation of climate andoceanographic models
science imaging — e.g., multispacecraft distributed interferometersflying in formation to enable imaging at microarcsecond resolution
Sandia National Labs MBARI AOSN NASA Terrestrial Planet Finder
J. Cortes MAE247 – Spring 2013
robotic networks decision making social networks
Engineered multi-agent systems
Embedded robotic systems and sensor networks for
high-stress, rapid deployment — e.g., disaster recovery networks
distributed environmental monitoring — e.g., portable chemicaland biological sensor arrays detecting toxic pollutants
autonomous sampling for biological applications — e.g.,monitoring of species in risk, validation of climate andoceanographic models
science imaging — e.g., multispacecraft distributed interferometersflying in formation to enable imaging at microarcsecond resolution
Sandia National Labs MBARI AOSN NASA Terrestrial Planet Finder
J. Cortes MAE247 – Spring 2013
sensor networks
self-organization
Further examples
Transportation networks: users that own part of the network makelocal decisions about the flow circulating over a portion of the network
Social networks: social agents and/or groups make decisions basedon local consensus or trends
Man-machine networks: humans make use of remote dynamicmachines while interacting over networks
Pervasive computing Ground traffic networks The Internet “Smart” power grids
J. Cortes MAE247 – Spring 2013
pervasive computing
Further examples
Transportation networks: users that own part of the network makelocal decisions about the flow circulating over a portion of the network
Social networks: social agents and/or groups make decisions basedon local consensus or trends
Man-machine networks: humans make use of remote dynamicmachines while interacting over networks
Pervasive computing Ground traffic networks The Internet “Smart” power grids
J. Cortes MAE247 – Spring 2013
traffic networks smart power grids
3 / 22
My main application of interest – the power grid
NASA Goddard Space Flight Center
Electric energy is critical forour technological civilization
Energy supply via power grid
Complexities: multiple scales,nonlinear, & non-local
4 / 22
Paradigm shifts in the operation of power networks
Traditional top to bottom operation:
I generate/transmit/distribute power
I hierarchical control & operation
Smart & green power to the people:
I distributed generation & deregulation
I demand response & load control
5 / 22
Challenges & opportunities in tomorrow’s power grid
www.offthegridnews.com
1 increasing renewables & deregulation
2 growing demand & operation at capacity
⇒ increasing volatility & complexity,decreasing robustness margins
Rapid technological and scientific advances:
1 re-instrumentation: sensors & actuators
2 complex & cyber-physical systems
⇒ cyber-coordination layer for smarter grids
6 / 22
Outline
Introduction
Complex network dynamicsSynchronizationVoltage collapse
Distributed decision makingMicrogridsWide-area control
Conclusions
Modeling: a power grid is a circuit
1 AC circuit with harmonicwaveforms Ei cos(θi + ωt)
2 active and reactive power flows
3 loads demanding constantactive and reactive power
4 synchronous generators& power electronic inverters
5 coupling via Kirchhoff & Ohm
Gij + i Biji j
Pi + i Qi
i
mech.torque
electr.torque
injection =∑
power flows
I active power: Pi =∑
j BijEiEj sin(θi − θj) + GijEiEj cos(θi − θj)I reactive power: Qi = −∑j BijEiEj cos(θi − θj) + GijEiEj sin(θi − θj)
7 / 22
complex network dynamics:
synchronization
Synchronization in power networkssync is crucial for AC power grids – a coupled oscillator analogy
sync is a trade-off
i(t)
weak coupling & heterogeneous
i(t)
strong coupling & homogeneous8 / 22
Synchronization in power networkssync is crucial for AC power grids – a coupled oscillator analogy
sync is a trade-off
i(t)
weak coupling & heterogeneous Blackout India July 30/31 2012 8 / 22
Our research: quantitative sync tests in complex networks
Sync cond’: (ntwk coupling)∩ (transfer capacity)> (heterogeneity)
θ(t)
θ(t)
220
309
310
120
103
209
102102
118
307
302
216
202
+ 0.1% load
sync cond’violated . . .
Reliability Test System 96 two loading conditions9 / 22
Our research: quantitative sync tests in complex networks
Sync cond’: (ntwk coupling)∩ (transfer capacity)> (heterogeneity)
θ(t)
θ(t)
220
309
310
120
103
209
102102
118
307
302
216
202
θ(t)
θ(t)
+ 0.1% load
Reliability Test System 96 two loading conditions
Ongoing work & next steps:
I analysis: sharper results for more detailed models
I analysis to design: hybrid control & remedial actions
9 / 22
complex network dynamics:
voltage collapse
Voltage collapse in power networks
reactive power instability: loading > capacity ⇒ voltages drop
recent outages: Quebec ’96, Northeast ’03, Scandinavia ’03, Athens ’04
“Voltage collapse is still
the biggest single threat
to the transmission sys-
tem. It’s what keeps me
awake at night.”
– Phil Harris, CEO PJM.
10 / 22
Voltage collapse on the back of an envelope
reactive power balance at load:
voltage
Esource
Eload
B
Qload
(fixed)
(variable)
EloadEsource0
Qload****
reactivepower
Q
load
= B E
load
(Eload
E
source
)
∃ high load voltage solution ⇔ (load) < (network)(source voltage)2/4
11 / 22
Our research: extending this intuition to complex networks
IEEE 39 bus system (New England)
Ongoing work & next steps:
existence & collapse cond’: (load) < (network)(source voltage)2/4
analysis to design: reactive compensation & renewable integration12 / 22
distributed decision making:
plug’n’play control inmicrogrids
Microgrids
StructureI low-voltage distribution networks
I grid-connected or islanded
I autonomously managed
ApplicationsI hospitals, military, campuses, large
vehicles, & isolated communities
BenefitsI naturally distributed for renewables
I flexible, efficient, & reliable
Operational challengesI volatile dynamics & low inertia
I plug’n’play & no central authority13 / 22
Conventional control architecture from bulk power ntwks
3. Tertiary control (offline)
Goal: optimize operation
Strategy: centralized & forecast
2. Secondary control (slower)
Goal: maintain operating point
Strategy: centralized
1. Primary control (fast)
Goal: stabilization & load sharing
Strategy: decentralized
Microgrids: distributed, model-free,online & without time-scale separation
⇒ break vertical & horizontal hierarchy
14 / 22
Plug’n’play architectureflat hierarchy, distributed, no time-scale separations, & model-free
Microgrid
…
…
……
…
…
source # 1 source # 2 source # n
Secondary
Primary
Tertiary
Secondary
Primary
Tertiary
Secondary
Primary
Tertiary
15 / 22
Plug’n’play architectureflat hierarchy, distributed, no time-scale separations, & model-free
Microgrid:physics& power flow
Diθi =P ∗i − Pi − Ωi
kiΩi =Diθi−∑
j ⊆ inverters
aij ·(
Ωi
Di− Ωj
Dj
)Di ∝ 1/αi
τiEi =−CiEi(Ei − E∗i ) − Qi − ei
κiei =−∑
j ⊆ inverters
aij ·(
Qi
Qi
− Qj
Qj
)−εei
Primary control:mimic oscillators
Tertiary control:marginal costs ∝ gains
Secondary control:diffusive averagingof injections
Ωi/Di
Qi EiθiPi
Qi/Qi
Qi/Qi
. . .
. . .
Ωi/Di
. . .
. . .
Ωk/Dk
Qk/Qk
Qj/Qj
Ωj/Dj
Pi =∑
jBijEiEj sin(θi − θj) + GijEiEj cos(θi − θj)
Qi = −∑
jBijEiEjcos(θi − θj) + GijEiEj sin(θi − θj)
source # i15 / 22
Experimental validation of control & opt. algorithmsin collaboration with microgrid research program @ University of Aalborg
DC S
ourc
e
LCL
filte
r
DC S
ourc
e
LCL
filte
r
DC S
ourc
e
LCL
filte
r
4DG
DC S
ourc
e
LCL
filte
r
1DG
2DG 3DG
Load 1 Load 2
12Z
23Z
34Z
1Z 2Z
0 10 20 30 40 50300
305
310
315
320
325
330Voltage Magni tudes
Time (s)
Voltage(V
)
0 10 20 30 40 50100
150
200
250
300
350
400
450
500Reactive Power Injections
Time (s)
Power(V
AR)
0 10 20 30 40 5049.5
49.6
49.7
49.8
49.9
50
50.1Voltage Frequency
Time (s)
Fre
quency(H
z)
0 10 20 30 40 50200
400
600
800
1000
1200A ct ive Power Injection
Time (s)
Power(W
)
t = 22s: load # 2
unplugged
t = 36s: load # 2
plugged back
t ∈ [0s, 7s]: primary
& tertiary control
t = 7s: secondary
control activated
16 / 22
Experimental validation of control & opt. algorithmsin collaboration with microgrid research program @ University of Aalborg
DC S
ourc
e
LCL
filte
r
DC S
ourc
e
LCL
filte
r
DC S
ourc
e
LCL
filte
r
4DG
DC S
ourc
e
LCL
filte
r
1DG
2DG 3DG
Load 1 Load 2
12Z
23Z
34Z
1Z 2Z
0 10 20 30 40 50300
305
310
315
320
325
330Voltage Magni tudes
Time (s)
Voltage(V
)
0 10 20 30 40 50100
150
200
250
300
350
400
450
500Reactive Power Injections
Time (s)
Power(V
AR)
0 10 20 30 40 5049.5
49.6
49.7
49.8
49.9
50
50.1Voltage Frequency
Time (s)
Fre
quency(H
z)
0 10 20 30 40 50200
400
600
800
1000
1200A ct ive Power Injection
Time (s)
Power(W
)
t = 22s: load # 2
unplugged
t = 36s: load # 2
plugged back
t ∈ [0s, 7s]: primary
& tertiary control
t = 7s: secondary
control activated
Ongoing work & next steps:
I time-domain modeling & control design
I integrate market/load dynamics & control
16 / 22
distributed decision making:
wide-area control
Inter-area oscillations in power networks
Blackout of August 10, 1996, resulted from instability of the 0.25 Hz mode
10
1
2
34
5
6
7
8
9
1112
13
14
15
16
South ArizonaSoCal
NoCal
PacNW
Canada
North
Montana
Utah
Source: http://certs.lbl.gov
0.25 Hz
17 / 22
Remedies against inter-area oscillationsconventional control
Physical layer: interconnected generators
Fully decentralized control:
effective against local oscillations
ineffective against inter-area oscillations
18 / 22
Remedies against inter-area oscillationswide-area control
Physical layer
Fully decentralized control
Distributed wide-area control
identification of architecture? sparse control design? optimality?18 / 22
Trade-off: control performance vs sparsity of architecture
K (γ) = arg minK
(J(K ) + γ · card(K )
)
optimal control = closed-loop performance + γ · sparse architectureper
form
ance
19 / 22
Case Study: IEEE 39 New England Power Gridsingle wide-area control link =⇒ nearly centralized performance
15
512
1110
7
8
9
4
3
1
2
17
18
14
16
19
20
21
24
26
27
28
31
32
34 33
36
38
39 22
35
6
13
30
37
25
29
23
1
10
8
2
3
6
9
4
7
5
F
Fig. 9. The New England test system [10], [11]. The system includes10 synchronous generators and 39 buses. Most of the buses have constantactive and reactive power loads. Coupled swing dynamics of 10 generatorsare studied in the case that a line-to-ground fault occurs at point F near bus16.
test system can be represented by
!i = "i,Hi
#fs"i = !Di"i + Pmi ! GiiE
2i !
10!
j=1,j !=i
EiEj ·
· Gij cos(!i ! !j) + Bij sin(!i ! !j),
"##$##%
(11)
where i = 2, . . . , 10. !i is the rotor angle of generator i withrespect to bus 1, and "i the rotor speed deviation of generatori relative to system angular frequency (2#fs = 2# " 60Hz).!1 is constant for the above assumption. The parametersfs, Hi, Pmi, Di, Ei, Gii, Gij , and Bij are in per unitsystem except for Hi and Di in second, and for fs in Helz.The mechanical input power Pmi to generator i and themagnitude Ei of internal voltage in generator i are assumedto be constant for transient stability studies [1], [2]. Hi isthe inertia constant of generator i, Di its damping coefficient,and they are constant. Gii is the internal conductance, andGij + jBij the transfer impedance between generators iand j; They are the parameters which change with networktopology changes. Note that electrical loads in the test systemare modeled as passive impedance [11].
B. Numerical Experiment
Coupled swing dynamics of 10 generators in thetest system are simulated. Ei and the initial condition(!i(0),"i(0) = 0) for generator i are fixed through powerflow calculation. Hi is fixed at the original values in [11].Pmi and constant power loads are assumed to be 50% at theirratings [22]. The damping Di is 0.005 s for all generators.Gii, Gij , and Bij are also based on the original line datain [11] and the power flow calculation. It is assumed thatthe test system is in a steady operating condition at t = 0 s,that a line-to-ground fault occurs at point F near bus 16 att = 1 s!20/(60Hz), and that line 16–17 trips at t = 1 s. Thefault duration is 20 cycles of a 60-Hz sine wave. The faultis simulated by adding a small impedance (10"7j) betweenbus 16 and ground. Fig. 10 shows coupled swings of rotorangle !i in the test system. The figure indicates that all rotorangles start to grow coherently at about 8 s. The coherentgrowing is global instability.
C. Remarks
It was confirmed that the system (11) in the New Eng-land test system shows global instability. A few comments
0 2 4 6 8 10-5
0
5
10
15
!i /
ra
d
10
02
03
04
05
0 2 4 6 8 10-5
0
5
10
15
!i /
ra
d
TIME / s
06
07
08
09
Fig. 10. Coupled swing of phase angle !i in New England test system.The fault duration is 20 cycles of a 60-Hz sine wave. The result is obtainedby numerical integration of eqs. (11).
are provided to discuss whether the instability in Fig. 10occurs in the corresponding real power system. First, theclassical model with constant voltage behind impedance isused for first swing criterion of transient stability [1]. This isbecause second and multi swings may be affected by voltagefluctuations, damping effects, controllers such as AVR, PSS,and governor. Second, the fault durations, which we fixed at20 cycles, are normally less than 10 cycles. Last, the loadcondition used above is different from the original one in[11]. We cannot hence argue that global instability occurs inthe real system. Analysis, however, does show a possibilityof global instability in real power systems.
IV. TOWARDS A CONTROL FOR GLOBAL SWING
INSTABILITY
Global instability is related to the undesirable phenomenonthat should be avoided by control. We introduce a keymechanism for the control problem and discuss controlstrategies for preventing or avoiding the instability.
A. Internal Resonance as Another Mechanism
Inspired by [12], we here describe the global instabilitywith dynamical systems theory close to internal resonance[23], [24]. Consider collective dynamics in the system (5).For the system (5) with small parameters pm and b, the set(!,") # S1 " R | " = 0 of states in the phase plane iscalled resonant surface [23], and its neighborhood resonantband. The phase plane is decomposed into the two parts:resonant band and high-energy zone outside of it. Here theinitial conditions of local and mode disturbances in Sec. IIindeed exist inside the resonant band. The collective motionbefore the onset of coherent growing is trapped near theresonant band. On the other hand, after the coherent growing,it escapes from the resonant band as shown in Figs. 3(b),4(b), 5, and 8(b) and (c). The trapped motion is almostintegrable and is regarded as a captured state in resonance[23]. At a moment, the integrable motion may be interruptedby small kicks that happen during the resonant band. That is,the so-called release from resonance [23] happens, and thecollective motion crosses the homoclinic orbit in Figs. 3(b),4(b), 5, and 8(b) and (c), and hence it goes away fromthe resonant band. It is therefore said that global instability
!"#$%&'''%()(*%(+,-.,*%/012-3*%)0-4%5677*%899: !"#$%&'
(')$
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 10, 2009 at 14:48 from IEEE Xplore. Restrictions apply.
1
10
Ongoing work & next steps:
cyber-physical security: corruption of wide-area signals
data-driven & learning: what if we don’t have a model?20 / 22
wrapping up
Summary & conclusions
Complex systems control
distributed, networks, & cyber-physical
Apps in power networks
complex network dynamics
distributed decision making
Surprisingly related apps
coordination of multi-robot networks
learning & agreement in social networks
and many others . . .
. . .
physical interaction
local subsystems and control
sensing & comm.
2
10
30 25
8
37
29
9
38
23
7
36
22
6
35
19
4
3320
5
34
10
3
32
6
2
31
1
8
7
5
4
3
18
17
26
27
28
24
21
16
1514
13
12
11
1
39
9
local system
local control
local system
local control
21 / 22
Acknowledgements
SynchronizationJohn Simpson-PorcoMisha ChertkovFrancesco BulloEnrique MalladaChanghong ZhaoMatthias Rungger
Voltage dynamicsMarco TodescatoBasilio GentileSandro Zampieri
Wide-area controlDiego RomeresMihailo JovanovicXiaofan Wu
MicrogridsQuobad ShafieeJosep GuerreroSairaj DhopleAbdullah HamadehBrian JohnsonJinxin ZhaoHedi Boattour
Robotic coordinationBruce Francis
Cyber-physical securityFabio Pasqualetti
Port-HamiltonianFrank AllgowerJorgen Johnsen
Social networksMihaela van der SchaarYuanzhang Xiao
...
Group @ ETH
Bala Kameshwar Poolla
plus some students onother prof’s payrolls . . .
more people to join . . .
22 / 22
thank you