Future Power Grids - DIgSILENT Americas · Future Power Grids 1 ... DGs, EV, security Complex...
Transcript of Future Power Grids - DIgSILENT Americas · Future Power Grids 1 ... DGs, EV, security Complex...
Future Networks Research at the University of Sydney6 Sep 2013
Future Power Grids 1
David Hill* and ZY Dong**
* Director of the Centre for Future Energy NetworksARC Professorial Fellow
**Head of School of Electrical & Information EngineeringThe University of Sydney
School of Electrical and Information
Engineering
Team: Profs Z-Y. Dong, Tony Vassallo, Jin Ma; Drs Gregor Verbic, Guo Chen; PhD students David Conroy, Hesam Marzooghi, Mehdi Garmroodi
School of EIE Industry Linkage
› EIE has been providing research and educational services in the following areas
› Advanced sensor technology
› Big data and computer networks
› Defence technology
› Fibre networks
› Next generation telco technologies
› Smart/future grid technologies
› Renewable energy
Current Research & Educational Capacity
2
EIE Research Matching of the IEEE / NIST SG Layered Architecture
NIST Conceptual Model
NIST System Architecture
3
Communications Architecture
Power System Architecture
IT/computing Architecture
Architecture Application AMI
Architecture Application PEV … Architecture
Application N
IP Smart Grid Architecture
Layer 1
Layer 2
Layer 3
Layer 4
4-layers smart grid architecture modelSource: IEEE/NIST
EIE
has
indi
vidu
al te
chno
logi
es &
R&
D c
apac
ity
Research & tools
Power System / market modeling, operations and planning
Optimization• Optimal Power Flow• System planning & operations• Economic Dispatch / Unit C’t• ATC Calculation• Wind farm optimisation
Monitoring• Load modeling and
parameter estimation• Contingences
Power system stability / vulnerability • Transient, Voltage Small-signal/osc. stab.• RE / DG/EV impact
Control Design• Power System Stabilizer• FACTS Device• Wind power/PV control
Soft Computing & Data Mining
• Computational intelligence• Learning • Multi-agents• Data mining
Computing Platform• HPC Server / Cluster• Parallel Computing • Grid Computing• Cloud computing
Software ToolsPSS/E,DSA,DigSILENT, PSCAD /EMTDC, GridLAB-D, ASPEN, SINCAL, OpenDSS, Prophet, Plexos, OPNET, NS-2/3
Smart Grid• Micro-grid• Grid Apps, Energy storage, Peakdemand, DGs, EV, security
Complex Systems• Vulnerability, security, topological analysis• Power network, comm’s & internet, …
Future Grid
Cyber System• ICTnetwork• Network modelling, security / vulnerability
Market simulation, risk, forecasting
Power Engineering Lab
› ABB power technologies equipment
› Robotic arm
› $1m digital power system simulator
› Usage:
- Professional Training
- Teaching
- Research and
- Final year thesis project
5
Equipment
New Challenges Faced by Power System Planning
6
� Traditional power system planning only considers three main components.
� A number of new factors have to be taken into account in power system planning, which introduces significant challenges.
Market benefit , reliability
Investment efficiency
Problems in planning
7
1. The coordination of power system planning and natural gas network planning.
2. Flexible/stochastic/hybrid methods for handling the uncertainties in the planning process.
3. Analyze the impacts of increasing customer participation (e.g. demand response and PHEVs)
Major question - resilience
› HOW to shape the whole grid so that:
- It adapts to the energy mix trajectory over decades
- with minimum cost
- specified stability margins
- limited carbon-emissions
- limited reliability
› We need to allow for controls used, including DM, reliability – orresults too conservative
› Excessive reliability costs too much – use probabilistic measures
8
Distributed architectures for demand response
9
Ref: Calloway and Hiskens, Proc IEEE, 2011
Grid2050 Architecture (Bakken et al.)
10
11
Bigger question - grid structure?
� There are already extreme scenarios
1. What if consumer prices go so high and local storage cost so low that most consumers can go off grid
2. Supergrids connecting continents to major sources of renewable power, e.g. Desertec
Studies of future grids have been too driven by possible ICT developments.
Economics will decide, e.g. cost of storage vs grid costs!
Australian Transmission Network
12
?
Zero Carbon Australia
› Assume:
- generation is renewable (rooftop to farms)
- fixed distributed storage (batteries to pumped storage)
- mobile storage (EVs)
› Aim: Provide an effective and efficient grid
› Real example: Masdar City, Abu Dabhi
13
14
15
Planning questions
› Can we continue incrementally?
› Can we keep transmission following generation, i.e. create (government controlled) ‘electricity highways’ or ‘super grids’ and hubs based on solar, wind etc
› Need expanded computer analysis to solve ‘network of network’ problems, e.g. how electricity and gas systems interact, and water, transport
› Must always model stability and capability to control
Stability challenges
› Robustness to all the uncertainty (Hiskens)
› Dependence on structure- Find the vulnerable points for collapse
- Study motifs for the new situations (project in Aus)
- Bigger questions: Backbone networks vs weakly connected clusters for diverse generation
› For theorists: How to guarantee stability from local checks - certificates (with some exchange)
- can these be granulated?
17
Renewables - case conclusions
› At this point just have a lot of observations
› All stability types affected by new dynamics, but it is hard to make many general statements
› Locations and generation types important, loads as before
› Structure of network important
Need more fundamental studies that can be applied generically
- Mechanism models, cf. as for voltage stability
- Control responses
18
19
Computational questions
› Numerous scenarios (millions)
› How to scale up solutions to such large systems and all scenarios – ideas from optimization and learning theory
› How to find the weak spots
- Transient instability
- Oscillations
- Vulnerable points for collapse
Australia’s FG programme
20
Funded by CSIRO, led by The University of Sydney - also a FG Forum for utilities on a shorter time-scale
Australia in 2050 (Ref: J.Sligar)
› Population 35million
› Large nuclear, geothermal and gas units with characteristics similar to present large units
› Thousands of small renewable generators assembled (virtual generators) at 25kv to 330kv
› Our special ‘stringy’ grid will be less so, but how?
› Still ‘no infinite bus’? cf. Denmark
› Major changes to loads: Al plants gone, new desalination plants, demand management, EVs
22
23
24
Planning interactions
25
26
Overseas
› USA DOE Future Grid university cluster
- http://www.pserc.wisc.edu/research/FutureGrid.aspx
› Fraunhofer Institut
- http://www.iwes.fraunhofer.de/en.html
› UK Grand Challenges
Outline
MotivationFuture gridsStability, control, planningResearch questions
Projects• Modelling with DM• Stability with renewables• Networked RE farm control• Probabilistiic margins• Stochastic TNEP• Generic market constraints with RE• Co-optimisation across GenTrans• Network vulnerability
Conclusions
27
Possible agenda for modelling
› Identify the granulated graphs
› Taxonomies of buses, lines, ‘motifs’
› Taxonomies of dynamics
› Aggregation up the levels
Lots of work here before we can begin analysis!
28
Simplied 14-generator model
› Areas
1. Snowy Mountains (hydro)
2. NSW
3. Victoria
4. Queensland
5. South Australia
29
Zero Carbon Australia FG
Source: Australian Sustainable Energy Zero Carbon Australia Stationary Energy Plan 2010
30
› Zero Carbon Australia proposal for HV grid upgrade
Tools for scenarios
› Scanning tool – fast simulation
› Revive direct methods, but load models no longer work
› Maybe EEAC type
› Most likely new algorithmic approaches
› Find weak points
› Network science ideas on vulnerability
31
Conclusions
› ‘Future grids’ a different and bigger question to ‘smart grids’ – long-term planning vs just better control
› Should we plan or grow the network incrementally – different scenarios
› Coordinated planning and control for high RE needs development
› Algorithms for planning (and control) – computational challenges
› Networks science ideas look promising BUT only after redone on useful models
32
33
System security and vulnerability assessment
› System security assessment to avoid cascading failure requires high computational efficiency- Deterministic vs probabilistic stability assessment
- Grid computing
- Load modelling & its impact
› In addition to conventional time domain methods and energy based methods, new methods have been proposed- Data mining & intelligent system & WAMS based approach
- Complex system based approach
- Sensitivity based approach
EPRI funds, IEEE Taskforce member on cascading failure, State Grid funds, PM&C funds
Intelligent system based security assessment
Knowledge Base
Generation
Input and Output
variables
Significant-
Feature selection
Intelligent
Algorithm training
Results utilization
Robustness
Reliability
Accuracy
.
.
.
.
.
.
Development Implementation
Intelligent Stability
Assessment
System
...
ZY Dong, P Zhang, Y Xu and KP Wong, “intelligent system for power system security assessment”, invited paper IEEE intelligentsystems journal (2010)
36
Complex Systemsapproach
Vertex (Bus)
undirected weighted graph
Vehicles((((vertex))))Road (Edge)
IEEE 14 bus system and the corresponding topologica l model
Edge (Branch)
The information exchange efficiency measures the network security
Load Modelling (load model parameter identification & generalisation)
› Research Problems:- Different values of parameters describe different dynamic properties of load
model. - using different dynamic response data in the task of parameter identification will
obtain different parameter values. - how the real dynamic properties of load model can be reflected by the appropriate
selection of load model parameters.- Specific measurement based load modelling, PSS_E, DigSILENT
› Support: EPRI, ARC, HKPU, SG/EPRI, Western Power Corp, AEMO› CIGRE C4.605: Modelling and Aggregation of Loads in Flexible Power Networks
On-line automatic var predictive control (EPRI)
› Operational data (CIM format)
› Predict system Var & recommend reactive power switching to maintain overall grid voltage stability
38
0 20 40 60 80 100 1200.9
0.95
1
1.05
1.1
0 20 40 60 80 100 1200.9
0.95
1
1.05
1.1
0 20 40 60 80 100 1200.9
0.95
1
1.05
1.1
0 20 40 60 80 100 1200.9
0.95
1
1.05
1.1
0 20 40 60 80 100 1200.9
0.95
1
1.05
1.1
Bus No.
V (p.
u.)
Initial Voltage profile
Resulting Voltage profile
39
Wind Resource Analysis and Prediction Package – OptiWIND©
› Short term wind forecast (min, hours, days)
- For operations
› Long term wind resource analysis (n x years)
- For wind farm planning, and wind Atlas
› Stand alone at client PC
› Server based
› Interactive graphic user interface
› Most commonly used wind turbines (Nordex, Vestas, SWT, GE)
Identified
offshore suitable
regions for wind
energy in HK
Map of the cumulative tracks of all tropical
cyclones during 1985–2005
A wind turbine
destroyed by
typhoon in Taiwan
40Wind turbine selection
(HK) Wind Speed Forecast
0 0.5 1 1.5 2 2 .5 3 3 .5 4 4.5
x 104
0
5
10
15
20
25
30
35
40
WindSpeed
Time (Minutes)
- Predict- Actual
Severe Tropical Storm Kammuri (Julian)
0 0 . 5 1 1 . 5 2 2 . 5 3 3 . 5 4 4 . 5
x 1 04
0
2
4
6
8
1 0
1 2
1 4
1 6
460 480 500 520 540 560 580 600
500
1000
1500
2000
2500
3000
Time (5 mins)
Win
d P
ow
er (
MW
)
95 % PI of Linear Regression
Observed Wind PowerLower Bound of PIUpper Bound of PI
�Short-term wind power interval forecasting (TAS wind)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20
5
10
15
20
25
30
35
40
45
Vestas V80 (2000 KW) Mean Power Production = 0.2029 (MW)
Capacity (MW)
Fre
quen
cy (
%)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.60
5
10
15
20
25
30
35
40
45
SWT-3.6-107 (3.6 MW) Mean Power Production = 0.3653 (MW)
Capacity (MW)
Fre
quen
cy (
%)
Wind Power Output: Vestas V80 vs Siemens SWT-3.6
-1000 -800 -600 -400 -200 0 200 400 600 800 1000-1000
-800
-600
-400
-200
0
200
400
600
800
1000
-1000 -800 -600 -400 -200 0 200 400 600 800 1000-1000
-800
-600
-400
-200
0
200
400
600
800
1000
-1000 -800 -600 -400 -200 0 200 400 600 800 1000-1000
-800
-600
-400
-200
0
200
400
600
800
1000
41
Wind Farm LayoutOptimisation
Mean Power Output = 8.1751 MWMean Power Losses = 0.4628 MW
Mean Power Output = 8.2034 MWMean Power Losses = 0.4345 MW
Mean Power Output = 8.2604 MWMean Power Losses = 0.3775 MW
Mean Power Output = 7.8700 MWMean Power Losses = 0.7679 MW
Mean Power Output = 7.9883 MWMean Power Losses = 0.6496 MW
Mean Power Output = 8.0725 MWMean Power Losses = 0.5654 MW
-1000 -800 -600 -400 -200 0 200 400 600 800 1000-1000
-800
-600
-400
-200
0
200
400
600
800
1000
-1000 -800 -600 -400 -200 0 200 400 600 800 1000-1000
-800
-600
-400
-200
0
200
400
600
800
1000
-1000 -800 -600 -400 -200 0 200 400 600 800 1000-1000
-800
-600
-400
-200
0
200
400
600
800
1000
11-Aug-2012 12-Aug-2012 13-Aug-2012 14-Aug-2012 15-Aug-2012 16-Aug-2012 17-Aug-2012 18-Aug-2012 19-Aug-20120
20
40
60
80
100
120
140
160
180
200kWh per 30 minute interval on 11/08/2012 to 18/08/2012
Date
kWh
PIPV
Solar PV› PV panel coating technology:
- Improve efficiency upto 20%
- Absorbs light from all angles
- Self-cleaning, scratch resistant
- Low maintenance costs
› PV impact studies (Ausgrid)- Peak demand
- Protection
- Energy storage
Microgrid› How to schedule a microgrid to
maximise the utilisation of RE while maintaining reliability
› point of common coupling (PCC) power, electric heater and boiler to accommodate the fluctuation of wind power
(in collaboration with Denmark)
CIENCentre for
Intelligent
Electricity
Networks
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5-80
-60
-40
-20
0
20
40
60
80
Time (s)
Rot
or a
ngle
(de
gree
)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6-200
-100
0
100
200
300
400
500
Time (s)
Rot
or a
ngle
(de
gree
)
Future smart grid: a social-cyber-physical system
Smart grid (cyber-physical system) simulation and security assessment tools
Physical (power system) ���� Cyber (ICT) and physical system ����social –cyber-physical system
4800 5000 5200 5400 5600 5800
50
100
150
200
250
Time of day (5 mins)
RR
P (
$ | M
Wh)
RRPforecast RRP
3000 4000 5000 6000 7000 8000
10
20
30
40
50
60
70
Time unit (5 mins)
Reg
iona
l Refer
ence
Pric
e ($
| MW
h) Real RRPForecasted RRP
Market Modelling
2000 2500 3000 3500 4000
10
20
30
40
50
60
70
80
Time unit (5 mins)
Reg
iona
l Ref
eren
ce P
rice
($ | M
Wh) Real RRP
Forecasted RRP
% Time
Pric
e ($
)
SRMC
Price Duration Curve Displaying Premium Calculation
δµ 2±=vP
Feature selection
-200.00
0.00
200.00
400.00
600.00
800.00
1,000.00
1,200.00
Sta
nwel
l
CS
QP
TC
Tar
ong
Mac
quar
ie
Ear
ing
Del
ta
Red
bank
Sno
wy
SE
CV
AG
L
Ene
rgy
Loy
Yan
g
Org
in
NR
G
TR
U
Hyd
ro
QLD NSW SNOWY VIC SA TAS
$Mill
ion
Free permit ETS purchase Non-ETS
0 5 10 15 20 25 30 35 40 45 500
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 5 10 15 20 25 30 35 40 45 500
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
NTL
CIENCentre for
Intelligent
Electricity
Networks
Future grid: National vs international? Transmission vs distribution? Energy, water, gas, telco, traffic, service, social, cash flow?
Thank [email protected]