Presentation PowerPoint

Dr. Pramod Khargonekar
University of Florida
Smart Grid and Integration of
Renewable Energy
Pramod P. Khargonekar
Department of Electrical and Computer Engineering
Impact of the Southeast in
the World's Renewable Energy Future
SEC Symposium
Atlanta, GA
February 12, 2013
Joint work with E. Bitar (Cornell), K. Poolla & P. Varaiya
(Berkeley), E. Bayenes (Valladolid), R. Rajgopal
(Stanford), M. Fang (Florida), …
Support from NSF and Florida Energy Systems
Electric Grid - Background
Smart Grid
Renewable Generation
Grid Integration
Future Directions
Focus on system operations,
not on specific hardware technologies
Electric Grid Characteristics
High voltage transmission network is a mesh network
Distribution networks are largely radial networks
(Socio-economic-technical) Dynamic system with phenomena at many timescales
Milliseconds, seconds, minutes, hours, days, months, and years
Geographically distributed yet tightly interconnected
Electric energy storage is very expensive and nearly impossible
Energy produced must equal energy consumed on a second-by-second basis
– power balance
A complex hierarchical distributed control system has evolved
over the years to ensure stability and performance of the large
scale networked power system
Power Balance
Power balance – balance power generation and consumption
on a second-by-second basis
Main Approach: adjust supply to meet demand with reliability
Natural uncertainty in consumption [load]
Use of reserve capacity to manage uncertainty and
Day-ahead – hourly schedules, one day ahead
Real-time – 5 minute schedules, 15 minutes ahead
Automatic generation control using system frequency
Deregulation of the electricity sector – unique mix of
engineering and economics
Smart Grid =
Power Grid + Sensors +
+ Computation + Control
Source: DOE
Smart Grid is a Vision
“a stronger, smarter, more efficient electricity
infrastructure that will encourage growth in
renewable energy sources, empower consumers
to reduce their energy use, and lay the
foundation for sustained, long-term economic
Steven Chu, U.S. Energy Secretary, 2009
Renewable Electricity Production
Mismatch Between Production and Consumption
Missing Link: Transmission Capacity
Source: EPRI
US Wind Capacity
US PV Installation
Source: IREC USA
Renewable Electricity Generation
Source: EIA
Variability of Wind and Solar
Source: CAISO
Power output varies in all
time frames:
Intermittency, uncontrollability, and uncertainty - principal
causes of difficulty at the operational level in integration of wind
and solar into the grid.
Large ramps,
up and down,
pose particular
Source: Integration of Renewable
Resources at 20% RPS, CAISO, 2010
Variability – Three Distinct Issues
Uncertainty – reliable predictions of power output
are hard, particularly day ahead
Uncontrollability – power output cannot be
controlled as desired
Intermittency – even if we could predict perfectly,
the power output is inherently
Variable Generation captures all three aspects into
a single phrase
Capacity Credit
Regulatory guidelines require periodic assessment and multi-year planning to
ensure sufficient generation capacity to meet demand
LOLP - 1 day in 10 years criterion (p=0.9997)
Resource Adequacy (RA) requirements; capacity markets
Particularly challenging in deregulated markets
Capacity credit
Nameplate capacity fraction for meeting RA requirements
Several probabilistic analysis techniques for CC calculation
What is the capacity credit of VG (wind)?
~PJM, MISO – 13%, NYISO – 10% (summer), SPP – 10%, E.ON – 8%, …
Could be even less at deep penetration
What are the impacts on power system planning?
How much traditional generation can be displaced by VG?
What happens at deep penetration of VG?
Better forecasts are extremely valuable
Prediction error decreases with horizon
Botterud et al, 2010
Forecast of wind velocity or solar flux –
numerical weather prediction models
Conversion of wind velocity or solar flux
into electric power
Natural opportunity for nonlinear
estimation methods enabled by sensors
and communications
Event detection methodologies to predict
the timing and magnitude of ramp
Use of neural networks and statistical
forecasting methods
Renewable Integration
Many large scale studies have been conducted in the last few years
Eastern Wind Integration and Transmission
Study (EWITS)
Western Wind and Solar Integration Study (WWSIS)
CAISO Integration Studies
Integration of Variable Generation Task Force (IVGTF)
NREL Studies, European projects, …
General conclusions: With sufficient transmission capacity, we can go up to
20-30% renewable electricity with significant impacts on power systems
operations – implications for:
Markets, reserves, balancing areas, flexibility of production stack, fast ramping
resources, storage, adjustable demand, ….
This is the focus of our work
California Results
Large simulation studies to estimate the impact of 20% and 33% RPS
Load following reserves:
2,292 MW in 2006 - 3,207 MW in 2012 - 4,423 MW in 2020
Up regulation reserves
277 MW in 2006 - 512 MW in 2012 - 1,135 MW in 2020
Is there a more rigorous method for estimating the additional
Are there techniques to reduce the need for these additional
Will/should renewable producers be required to provide their own
Wind Power Model
Wind power is taken to be a discrete-time stochastic
Normalized to name-plate capacity:
CDF of the wind power process:
Time averaged CDF:
Market Model
Two settlement market
Contract, C
Wind power w
Forward day-ahead market
(q, l)
Real-time market
Wind producer offers a contract for constant power C in the
DAM at price p
Imbalance prices in the real-time market: q shortfall price, l
excess price
Imbalance prices q and l are taken to be random variables
while p is assumed to be known
Profit Function
Energy shortfall
Energy excess
Optimal Policy is a Quantile Policy
(1) Wind power producer is a price taker
(2) Wind power and prices are uncorrelated
Topic of intense research
Will have a major impact on
Renewables and electric transport
Source: Electrical Storage Association
Energy Storage Model
Stored energy dissipation - a
Injection and extraction efficiency – hinj, hext
Optimal Contract with Storage
Same two settlement market as before
At each time instant, we can choose to inject energy into
or extract energy from the storage device to maximize
the net profit:
S = net injection = Pinj - Pext
Storage operation policy = g
Admissible policies: all feedback policies that depend on
the past values of w and e
Lemma: Greedy policy, g*, is optimal.
Theorem: J(g*, C) is concave in C.
Theorem: Optimal profit is a concave, monotone nondecreasing function of the storage size .
Benefits of Aggregation
Consider a collection of geographically dispersed VG
Intuition: Averaging can reduce variability
Can a group of wind power producers increase their
collective profits by aggregating and offering their
power output as a single entity?
What profit sharing policy will ensure that the
producers cooperate?
Bayens et al. CDC’2011
Paradigm Change
Current: adjust the generation to meet random demand
Future: adjust demand to meet random generation
Flexible Demand: heating, air-conditioning, refrigeration, water heaters,
EVs, …
These are energy consumers, not power consumers
How can we optimize aggregate and optimize flexibility of large
numbers of individual flexible loads?
How can sensing and communications be used for distributed
control of flexible loads?
What incentive and pricing mechanisms will be effective in getting
consumers to participate in adjustable demand programs?
How can these distributed resources be integrated into power
system operations with large RG penetration?
Distributed Renewable Generation
Large numbers of solar, wind, CHP, and micro-generators in the
distribution system
Adjustable demand, electric vehicles
Sensing, communications, computing, control (SG)
What is the optimal, scalable, control and communications
architecture to control such a large scale distributed power system?
How can we do this while respecting the legacy centralized grid and
minimize the need for additional reserves?
What level of renewable penetration can be achieved in such a
distributed scenario?
GRIP: Grid with Intelligent Periphery
Coordinated aggregation & control using smart grid sensing,
communications, computation, and control
Bakken et al, SmartGridCom’2011
Our Publications
 E. Bitar, A. Giani, R. Rajagopal, D. Varagnolo, P. P. Khargonekar, K. Poolla, P. P. Varaiya,
“Optimal Contracts for Wind Power Producers in Electricity Markets,” Proc. 50th IEEE
Conference on Decision and Control, pp. 1919-1926, 2010.
 E. Bitar, R. Rajagopal, P. P. Khargonekar, and K. Poolla, “Optimal Bidding Strategies for
Wind Power Producers: the Role of Reserve Margins and Energy Storage,” Proc.
American Control Conference, pp. , June 2011.
 E. Bitar, P. P. Khargonekar, and K. Poolla, “Systems and Control Opportunities in Smart
Grid and Renewable Integration,” Proc. 2011 IFAC World Congress, Milan, ITALY.
 D. Bakken, A. Bose, K. M. Chandy, P. P. Khargonekar, A. Kuh, S. Low, A. von Meier, K.
Poolla, P. P. Varaiya, and F. Wu, “GRIP – Grids with Intelligent Periphery: Control
Architectures for Grid2050,” Proc. IEEE SmartGridComm, 2011.
 A. Giani, E. Bitar, M. Garcia, M. McQueen, P. P. Khargonekar, and K. Poolla, “Smart Grid
Data Integrity Attacks: Characterizations and Countermeasures,” Proc. IEEE
SmartGridComm, 2011
 E. Baeyens, E. Bitar, P. P. Khargonekar, and K. Poolla, “Wind Energy Aggregation: A
Coalitional Game Approach,” Proc. IEEE Conference on Decision and Control, 2011.
 E. Bitar, K. Poolla, P. P. Khargonekar, R. Rajgopal, P. Varaiya, and F. Wu, “Selling
Random Energy,” Hawaii International Conference on Systems Science, 2012.
 E. Bitar, R. Rajagopal, P. P. Khargonekar, K. R. Poolla, and P. Varaiya, “Bringing Wind
Energy to Market,” submitted for publication to IEEE Transactions on Power Systems .
[email protected]
Dr. Pramod Khargonekar
University of Florida

similar documents