Autonomy-Oriented Mechanisms for Efficient Energy Distribution

1
Autonomy-Oriented Mechanisms for Efficient Energy
Distribution

• Presenter Benyun Shi
• Principal Supervisor Prof. Jiming Liu
• Co-Supervisor CHEUNG, William Kwok Wai
• Department of Computer Science
• 11th PGDay
• Mar. 15, 2010

2
Outline

• The Energy Distribution Problem
• World Energy Status
• Challenges
• Our research focus
• The Autonomy-Oriented Mechanism
• Four local behavior-based algorithms
• Preliminary Simulations
• Conclusion and Future Work

3
Energy Status (1)-- Uneven geographical
availability
75
Eurasia
56
Middle East
The Worlds Proved Oil Reserves
The Worlds Proved NG Reserves
Data from the Oil and Gas Journal and
International Energy Agency.
4
Energy Status (2) -- Imbalanced energy
utilization
(Mtoe)
Total primary energy supply of the world from
1971 to 2007. (Adopted from Key World Energy
Statistics 2009)
Worlds total energy use from1965 to 2008.
Data from the British Petroleum.
5
The General Energy Distribution Problems

• Efficiently, economically, and reliably
  distribute energy resources either worldwide or
  within a country/region.
• Issues (how to)
• Energy price e.g., electricity price in power
  grid 1
• Distribution infrastructure investment e.g.,
  pipelines, railways 2
• Cascading control or congestion management in
  Power grid 34
• Energy markets e.g., world oil market and oil
  futures market
• Logistics networks of energy resources 56
• Distribute energy from suppliers to consumers
  under certain physical constraints.
• Maintain reliable and secure energy distribution
  system
• Other issues

6
Challenges

• The energy distribution problems are complex in
  terms of
• Energy supply/demand may dynamically changing
• Endogenously increase by population or economy
  development
• Exogenously severe weather
• The relationships between energy suppliers and
  consumers may evolve over time
• The information may only be partially available
• Due to private issues or competitions
• Suppliers/consumers make decisions based on their
  own benefits

7
Related Work (1)

• Systems Dynamics Approach (or Macro-modeling)
• Issues
• Understand the relationships among different
  component in energy system
• Determine roles of energy system in social,
  economic, and environmental systems
• Simulate the real world
• Drawbacks
• Represent relationships based on statistical
  data
• Hard to represent dynamics, e.g., technology
  innovations
• Need exascale computing 7

Predictions can be done in macro-level
Hard to provide distribution solutions in
specific energy domains!
8
Related Work (2)

• Static Network Optimization Approach (or
  Micro-modeling)
• Optimize certain objective on static networks
• Mathematical Optimization
• E.g., U.S. integrated energy system 56
• E.g., Economic dispatch in natural gas networks
  8
• Dynamics-driven Network Optimization Approach
• Form optimal networks based on certain dynamics
  on the network
• The by nature open, distributed, and dynamically
  evolving energy distribution system need
  decentralized approaches

9
Our Research Focus

• Autonomy-Oriented Mechanism for Energy
  Distribution
• Entities
• Represent either suppliers or consumers, or a
  group of them
• Interactions
• Entities interact with each other as well as
  their environment to collect information
• Behavioral rules
• Exploration behavior
• Regulation behavior
• Objectives
• to characterize the underlying mechanisms of the
  energy distribution system through local
  interactions between entities with different
  behavioral rules
• to provide scalable distribution solutions

Self-Organized Mechanism
10
Starting with a Static Energy Distribution Problem

• A set of energy suppliers/consumers with constant
  energy supply/demand
• Assume that total supply total demand
• Represent energy distribution cost among energy
  suppliers and demander as a predefined matrix
  CMatrixnncij
• Symmetric cij cji
• Triangle inequality cij cjk ? cik
• Objectives
• To meet the consumers demand
• To minimize the global distribution cost
• Questions to be tackled in this work
• How does an optimal energy distribution network
  can energy through local dynamic of
  supplier/consumer entities?
• What kind of local behaviors are crucial for
  achieving final optimal energy distribution
  network?

11
Entities Behaviors

• Exploration Behavior (Self-avoiding random walk)
• Explore but do not memorize (Algorithm 1 and 2)
• Explore and memorize for future utilization
  (Algorithm 3 and 4)
• Regulation Behavior (decide whom to trade with)
• First come first serve rule (Algorithm 1)
• Competition (Algorithm 2 and 3)
• Proactively send request (Algorithm 4)

Hypothesis by memorizing information for future
utilization, it is much easier to find a path
with small distribution cost.
Hypothesis by proactively regulating trading
partners and sending requests, it is more likely
to find appropriate partners than passively
trading with visitors.
12
Four Algorithms

• Algorithm 1 Self-avoiding Random Walk with
  First-come-first-serve
• Explore but not memorize
• Algorithm 2 Self-avoiding Random Walk with
  Competition
• Explore but not memorize
• Algorithm 3 Self-avoiding Random Walk with
  Information Sharing
• Explore and memorize for future utilization
• Algorithm 4 Self-avoiding Random Walk with
  Information Passing
• Explore and memorize for future utilization
• Proactively regulate trading partners

13
Simulations -- Measurements

• Distribution Cost
• Global Cost of Energy Flow Network
• Per Unit Cost of Energy Flow Network
• Scalability
• When the number of suppliers/consumers increase,
  can the autonomy-oriented mechanism remain
  efficient?

Distributed quantity along link lij
14
Observations (1)
Validate hypothesis about exploring with
memory Validate hypothesis about proactively
regulating trading partners
15
Observations (2)
Scalability The per unit distribution cost of
energy flow network of Algorithm 4 approaches to
optimal solution as the number of nodes increase
form 10 to 1000.
16
Conclusions

• The autonomy-oriented mechanism study the energy
  distribution problem from a bottom-up viewpoint
• Global objectives can be approximately reached
  through local interactions of behavior-based
  autonomous entities
• Appropriate exploration and regulation behaviors
  play important roles
• Scalability makes it possible to deal with
  large-scale energy distribution problems, like
  smart grid.

17
Future Work

• It can be naturally extended to deal with open,
  distributed, as well as dynamically evolving
  energy distribution problems.
• How does the energy flow network evolve in an
  open, unpredictable energy distribution system?
• What kind of local dynamics between supplies and
  consumers can improve the robustness and
  stability of the energy distribution system?
• What kind of energy trading mechanism (market)
  can be formed? What are the critical factors
  for the stability of the market?

18
References

• 1 M. Bjørdal. Topics on Electricity
  Transmission Pricing. PhD thesis, Norwegian
  School of Economics and Business Administration,
  Bergen, 2000.
• 2 Oil Division. A compendium of electric
  reliability frameworks across canada. Technical
  report, Petroleum Resources Branch, Canada, 2008.
• 3 A. E. Motter. Cascade control and defense in
  complex networks. Physical Review Letters,
  93(9)098701, 2004.
• 4 F. Schweppe, M. Caramanis, R. Tabors, and R.
  Bohn. Spot Pricing of Electricity. Kluwer
  Academic Publishers, Norwell, Massachusetts,
  1988.
• 5 A. Quelhas, E. Gil, J. D. McCalley, and S.
  M. Ryan. A multiperiod generalized network flow
  model of the U.S. integrated energy system Part
  I - model description. IEEE Transaction on Power
  Systems, 22(2)829836, May 2007.
• 6 A. Quelhas and J. D. McCalley. A multiperiod
  generalized network flow model of the U.S.
  integrated energy system Part II - simulation
  results. IEEE Transaction on Power Systems,
  22(2)837844, May 2007.
• 7 H. Simon, et al. Modeling and simulation at
  the exascale for energy and the environment.
  Technical report, Report on the Advanced
  Scientific Computing Research Town Hall Meetings
  on Simulation and Modeling at the Exascale for
  Energy, Ecological Sustainability and Global
  Security (E3), 2007.
• 8 K. T. Midthun, M. Bjorndal, and A.
  Tomasgard. Modeling optimal economic dispatch and
  system effects in natural gas networks. The
  Energy Journal, 30(4), 2009.

19
(No Transcript)