Microsoft PowerPoint - j_nutaro [Compatibility Mode]

ntegrated information grid and electric grid modeling Algorithms for distributed response to Algorithms for distributed response to
contingencies and Data-analysis Hybrid continuous-discrete modeling and simulation Visualization support
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
2 Visualization support
Data credit:
http://www.nerc.com/dawg
, FERC, EIA. Plot credits: Prof. John Doyle, Prof. Massoud Amin. ORNL R&D: Interdependent
Components
Components
User
Feedback
Wide-area visualization
Spatio-temporal
Feedback
and Decision
Support
representation
Application
scenarios: grid-
Simulation-
based
Data-
l i
Large-Scale
state, outages
User-
discovery
Algorithm
parameters
discovery
Scenario-
steering
analysis
and
Algorithms
Modeling
and
Simulation
p
Specifications
Distributed control and
communication
Data directed disco er
Hybrid simulation
Parallel contingency evaluation
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
3
Data-directed discovery Computational Elements in the Grid
of the Future
Networked
actuators at
electrical
substations can
connect and
disconnected
loads in fixed
loads in fixed
increments
Monitor and
f
control software
attached to the
generators
control the
control the
actuators through
an IP-based,
wide area
t
k
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
4
network 1
Problem addressed
1
: accurately
simulating automatic emergency load
shedding and restoration in a smart
shedding and restoration in a smart
grid of the future
( a smart grid of the future will have network
enabled control to prevent power system
failures due to under-frequency events)
failures due to under frequency events)
1
James Nutaro, Phani Teja Kuruganti, Laurie Miller, Sara Mullen, Mallikarjun Shankar,
Integrated Hybrid Simulation of Electric Power and Communication Systems, IEEE Power
Engineering Society Tampa June 24 28 2007; To Appear
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
5
Engineering Society, Tampa, June 24-28, 2007; To Appear. Line frequency
Transd cer (a er big electric
Transducer (a very big electric
motor) converts mechanical
power to electrical power
Electrical demand
causes shaft to
decelerate
decelerate
Torque applied by the
power plant to the shaft
causes acceleration
An circuit breaker will
disconnect the power
plant from its load if the
causes acceleration
plant from its load if the
output signal frequency
deviates much from 60
Hz
End user uses electric
power, placing a load
on the accelerating
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
6
rotor via the
transducer Cause of frequency related failures
Power plants are large
Industrial loads can switch large
electrical loads on and off quickly
machines with a lot of
inertia they need time
to adjust the amount of
torque applied to the
Frequency changes when
there is difference between
electrical load
and
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
7
rotor
electrical load
and
mechanical power Objective of an under-frequency
Objective of an under-frequency
loading shedding system Problem: Prevent circuit breakers from
disconnecting due to the line frequency
exceeding 60 Hz by more than their safety
margin
S l ti
If th l
d i t
hi h f
t Solution: If the load is too high for too
long, then cut some of the load
Bring the cut load back on line as soon as Bring the cut load back on-line as soon as
possible!
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
8 IEEE 14 Bus Test Case - Structure
Generator
Load
El t i l b
Electrical bus
Transmission line
Transmission line
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
9 Continuous Dynamics
)
P
P
( =
&
1 Angular acceleration of the shaft
)
P
P
.
R
/
b
k
(
P
)
P
P
(
M
m
m
g
agc
m
e
m
+
+
+ =
&
&&
25
0
100 Control of the power plants
mechanical power production












l
lg
ll
m
Y
Y
P
r
r
Set of linear equations describing how
electric loads at the busses become










=




g
l
gg
gl
g
l
m
Y
Y
P
r
r
e ect c oads at t e busses beco e
electrical power demand at the
generator terminals
Mechanical
power and
electrical
demand at the
generator and
Admittance matrix
for the electrical
network that
connects
generators and
Bus and generator
angles relative to a
rotating reference
frame
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
10
generator and
loads
generators and
loads State event: Generator disconnect
Frequency
protection breaker
Frequency
protection breaker




lg
ll
Y
Y
closed
protection breaker
open
|60 Hz - | > 0.1 Hz
Generator output
frequency exceeds



lg
ll
Y
Y




gg
gl
g
Y
Y
frequency exceeds
safe threshold
Disconnect causes




gg
gl
g
Y
Y
an immediate
change in the
admittance matrix
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
11 Time event: Piecewise constant
Time event: Piecewise constant
loads
Changes in
electrical load are
modeled by
modeled by
relatively
instantaneous
events
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
12 Example of an under-frequency failure
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
13 Basic controller concept Monitor power production and demand at the
generator terminals and the generator angular
acceleration
acceleration Periodically calculate an estimate of time to
under-frequency failure
q
y If that time is too short, send a network message
asking actuators to shed a load increment
If th t ti
i
l
d
t
k If that time is very long, send a network message
telling actuators to reconnect a load increment (if
any are disconnected)
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
14
y
) Controllers are modeled as discrete
Controllers are modeled as discrete
event systems
O t t
t
t
k
Output events cause a network
message to be broadcasted to the
actuators is the controllers load shed request is the load shed increment
t
s
is the time until power demand equals production
t
f
is the time to failure threshold
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
15
k is a safety factor The communication network is
The communication network is
modeled as a discrete event system
Network communication
lines follow the electrical
transmission lines
Broadcasts are
implemented with a flooding
protocol
protocol
UDP/IP is used as the
communication protocol
A six hop route
from Tazwell to
Homer
communication protocol
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
16 Discrete event and continuous
Discrete event and continuous
subsystems are tightly coupled
Communication network
Control
Actuation
Actuation
Power generation
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
17
Electrical load Test scenario
Loads 7, 8, 10, 11,
, ,
,
,
and 12 disconnect
suddenly at t = 1
Restored suddenly
Restored suddenly
with some
additional load at t
= 10
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
18 Result without control
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
19 Experiments Vary base link latency and throughput in
the communication network Can the system survive? Does the system stabilize? How much load is shed by the controller?
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
20 130 milliseconds, 1 Mbps throughput
60.1
Genr. 1
60.1
Genr. 1
Genr. 2
Genr. 3
Genr. 4
Genr. 5
60.05
(H
z)
Frequency ripple caused by the control action
60
frequency
(H
59.95
59.9
0
2
4
6
8
10
12
14
16
18
20
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
21
0
2
4
6
8
10
12
14
16
18
20
time (seconds) 3.5
Load Service Frac. 1
130 milliseconds, 1 Mbps throughput
3
Load Service Frac. 1
Load Service Frac. 2
Load Service Frac. 3
Load Service Frac. 4
Load Service Frac. 5
Total Load Serviced
Total Load Demand
2
2.5
Oscillation in the control
Oscillations in the total
load due to
disconnect/reconnect
cycle of the controller
1.5
2
Oscillation in the control
signals caused by
message delays in the
network
cycle of the controller
1
0
0.5
0
2
4
6
8
10
12
14
16
18
20
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
22
0
2
4
6
8
10
12
14
16
18
20
time (seconds) 20 milliseconds, 256 Kbps
60.1
Genr. 1
Genr. 2
60.05
Genr. 2
Genr. 3
Genr. 4
Genr. 5
60.05
(H
z)
Frequency stabilizes quickly following the event
60
frequency
(H
59.95
59.9
0
2
4
6
8
10
12
14
16
18
20
O
AK
R
IDGE
N
ATIONAL
L
ABORATORY
U. S. D
EPARTMENT OF
E
NERGY
23
0
2
4
6
8
10
12
14
16
18
20
time (seconds) 3.5
Load Service Frac. 1
Load Service Frac