Kevin Tomsovic* and Mengstab Gebremicael

1
Modeling the Interaction between the Technical,
Social, Economic and Environmental Components of
Large Scale Electric Power Systems

• Kevin Tomsovic and Mengstab Gebremicael
• School or Electrical Engineering and Computer
  Science Washington State University
• Currently on leave at National Science
  Foundation

2
Outline of Presentation

• Some Observations and Questions
• When should we expect power plant construction
• Expansion of WSUs Work in Several Areas
• Collaboration in Research
• System Dynamics
• Overall Study Approach
• Study Bench Marks
• Actual Model in Simulink
• Conclusions and future works

3
Some Observations and Questions

• Deregulation has been met in every case by
  unintended consequences, some reaching the
  level of a crisis
• The electric power industry has historically gone
  through periods of boom and bust cycles.
• Is it the fundamental nature of generation
  investment and technology that leads to these
  cycles and crises?
• Do transmission system limits and reliability
  considerations exacerbate the difficulty of
  predicting policy outcomes?
• How are the cycles influenced by the new market
  policies?
• How do various incentives programs (e.g. capacity
  payment, tradable green certificates) impact the
  planning process?
• ? Can regulatory policies and new engineering
  approaches relieve these cycles and resulting
  societal costs?

4
Disparate System Views

• System dynamics research shows the tendency
  towards boom/bust cycles from generation
  investment and construction permit policies.
• Engineers understand the operating limits of the
  transmission system.
• Economists know market structures that generally
  lead to more efficient economic behavior from
  suppliers.
• Policy makers may set goals based on limited
  understanding of operations, e.g., 20 renewables
  in 20 years.
• ? But how do these areas interact?

5
When should we expect power plant construction
to appear?

• Just in time to cause the market to clear at an
  average annual price that matches the total cost
  of a new power plant?
• In waves of boom and bust?
• Textbook answer
• Construction will appear just in time to keep
  market prices at the cost of a new entrant.
• The answer from other industries (agriculture,
  mining, real estate)?

6
Construction will be in waves of boom and bust

Ref Land Values and Real Estat Construction in
Chicago traced from Hoyt (1933)
7
Lessons from Other Industries

• Pay attention to physical factors, such as long
  lead times for permitting and construction
• Include the behavioral factors, such as the
  tendency to discount the construction in progress
• Expect psychological factors to shape investor
  behavior and our discussion of boom bust

8
(No Transcript)
9
Boom and Bust in Power Systems
Price spikes reappear in 2007
Price Implications of Base Case Simulation from
Nov 2001
10
Expansion of WSUs Work in Several Areas

• Long term investment dynamics
• System security
• Existing models do not show impact of
  transmission systems
• System security in operations
• Transmission planning processes
• Market models and investment behavior
• Bidding behavior
• Impact of congestion on bidding behavior
• More sophisticated market rules
• New generation technologies (e.g., dispersed
  generation units)
• Environmental impacts

11
Research PlanDevelopment of models to provide
improved inputs to the system dynamics simulation

• Transmission systems
• Simplified network models appropriate for
  studying longer term trends with the inclusion of
  all important effects (regional bottlenecks,
  etc.). More detailed than simple reserves.
• Transmission planning processes under various
  economic structures.
• Markets
• Consider impact of market rules and supplier
  gaming
• Environmental Impact
• Role of renewable targets and other related
  policies

12

• Collaboration in Research

• Cooperation with West African Researchers
• Development of models appropriate for West
  African Power Pool
• Study impact of weakly meshed transmission
  systems
• Modifications for behavioral, regulatory and
  environmental differences
• Emphasis on Matlab models rather than Vensim

13
System Dynamics

• Originated by applying the concepts of feedback
  theory to the study of industrial systems
• Models are one of many tools to help in the study
  of chaos and complexity
• Models are constructed to help understand why
  patterns (growth, decay, and oscillations) occur
• Designed for general understanding, not point
  prediction
• Emphasizes high level intuitive construction of
  models.
• No explicit representation of dynamic equations.
• Awkward implementation of numerical algorithms
  (e.g., market clearing processes)

14
System DynamicsModeling Issues for Generation
Investment

• Investors expected prices several years ahead
• Time lags in construction of facilities
• Investor behavior (bounded rationality)
• External economic factors and other unknowns
• Reserve margin base decision
• More information at WSU Website
  http//www.wsu.edu/forda click research on boom
  and bust in the competitive electric industry

15
Software Issues and Model Development

• Engineering
• Emphasizes physical and precision of models at
  potentially the expense of higher level insights.
• Explicit representation of dynamics.
• Sophisticated computational methods.
• Variety of analytical methods
• Software tools (Matlab)
• Extensive libraries of computational tools
• Model building labor intensive calculations.

• System dynamics
• Emphasizes high level intuitive construction of
  models.
• No explicit representation of dynamic equations.
• Awkward implementation of numerical algorithms
  (e.g., market clearing processes)
• Software tools (Stella, Vensim)
• Powerful tools for scenario studies
• Fast methods to build models
• Not open to sophisticated numerical calculations

16
System Dynamics - For Engineering

• Simulink
• Is an interactive tool for modeling, simulating,
  and analyzing dynamic systems.
• Extensive libraries of computational tools
• Model building labor intensive
• Is an extension to MATLAB which uses an
  icon-driven interface for the construction of a
  block diagram representation of a process.
• The tool choice for control system design,
  signal processing, communication, and other
  simulation applications
• Explicit representation of dynamics.
• Sophisticated computational methods.
• Variety of analytical methods
• The block diagram represent the actual math
  (Different blocks for different math expressions)
• Emphasizes physical and precision of models at
  potentially the expense of higher level insights.

17
Simulink Model for Construction of New Combined
Cycle Plants
18
Overall Study Approach

• Follow some suggested steps of modeling
• Get acquainted with the system
• Be specific about the dynamic problem
• Draw the causal loop diagram
• Run the model to get the reference model
• Conduct sensitivity analysis
• Test the impact of policies
• Scenario analysis various assumptions
• Price forecasts
• Economic growth
• Weather variables
• Investor behavior variations
• Reserve margins
• Verification from historical data

19
Benchmark Systems - WECC

• Five regions
• North West Power Pool
• Rocky Mountain Power Area
• Arizona - New Mexico -Southern Nevada Power
• Northern California
• Southern California.
• Resources, load growth, and so on, vary by area
• No transmission constraints within regions
• Network parameters derived from DC network load
  flow model

20
Benchmark Systems WAPP (cont)14 Countries of
West Africa proposed West African Power Pool
21
Simulink Model for WAPP(RM base investment)
22
Simulink Model for WAPP (Price based investment)
23
Simulink Model S function Price and generation
computation

• Cost Function
• - linear marginal cost function (incremental
  cost)
• - Average full costs for will be an integral
  over the marginal cost function

24
S function (cont)

• The total cost is given by
• Matlab formulation of quadratic function, as
  expected by the quadprog function
• The linear terms (vector b) can also be expressed
  as

25
S function (cont)

• Network Constraints
• - DC load flow which
• relates injected nodal real powers, voltage phase
  angles and real power flows in network elements
  (branches).
• assumes that voltage magnitudes are all equal to
  1 p.u.
• assumes that the network is lossless (branch is
  represented only by its equivalent reactance )
• - The first set of equations relates injected
  nodal real powers and nodal voltage phase
  angles
• -The B' matrix is derived from the
  bus-admittance (inverse of bus-impedance)
• quadratic, symmetric, for a network with n nodes
  has dimension n-1 x n-1.
• bii sum of all inversed reactances of the
  branches connected to node i
• bij negative sum of all inversed reactances of
  the branches connected between nodes i and j

26
S function (cont)

• Network Constraints (cont)
• - The second set of equations relates
  real power flows in network branches Pflow and
  nodal voltage phase angles
• - To solve for power flows first find phase
  angles
• - The power flow equations can be written as
• - The inequality constraints imposed by the
  network elements capacities are

27
Simulation Results (price, generation)
28
Simulation Results (construction on display)
Construction (30 month simulation
29
Conclusions and Future work

• Engineering model Detailed studies
  Computationally intense
• Power flow model for transmission constraints
• Full market model with possibility of strategic
  bidding for energy and reserves
• Use of analytical models for investor behavior
  that lose some of their intuitive feel
• Investigation of stability analysis methods for
  developed models (initially using linearization
• As pointed out the model is for learning, and
  improved understanding of the interaction between
  technical, social, economical, and environmental
  factors in power plant investment
• So far the reference mode (boom and bust cycle)
  of construction is not attained
• Different test scenarios are going to be
  conducted
• Validation of the model using historical data is
  expected