Transcript New Opportunities for Control and Optimization

New Opportunities for Control and
Optimization in the Future Power
Industry Environment
Chen-Ching Liu
University of Washington
Outline
•
•
•
•
•
•
•
New Electricity Market Environment
Ancillary Service Selection
Optimal Bidding Decisions
Flexible Contract Pricing
Risk Management in a Competitive Market
Defense Plans
Control of Available Transfer Capabilities (ATC)
Electricity Market
Environment
ISO / RTO (managing the use of the grid,
coordinating the market)
Large Consumers
Spot Market
Generation
Companies
Bilateral
Contracts
Distribution
Companies
Other Consumers
Marketers
Ancillary Service (AS) Selection
• Objective: Min Cost = S(bid-pricequantity)
• Controls: Amount of AS cleared per bus, SC
• Constraints:
– System reserve and regulation requirements
– Max ramp rates
– Max and min bid block amounts
– Unit capacity limits
 Problem: How to make least-cost decisions for
AS selection given a set of AS bids?
Optimization Model
congestion
min 
V , iG
jD
kQ
ciG PiG  c Dj PjD  ckQ Qk 
(cires Pires   reseires Pires res )  (cireg Pireg  eireg Pireg reg )]
Subject to:
reserve
Unit capacity
regulation
Power Flow
System Reliability
Pi G  Pi res  0.5Pi reg  Pi max
Pi (V , )  Pi G  Pi D
Pi min  Pi G  0.5Pi reg
Qi (V , )  Q  Q
G
i
D
i
Bid Block Limits
0  Pi reg  Pi reg ,max
0  Pi
res
 Pi
reactive
power
res ,max
Qimin  QiG  Qimax
Pi min  Pi  Pi max , i  G  D
P
reg
 Re g _ Re q
P
res
 Re s _ Re q
iRe g
iRe s
Security
V max  V  V min
Pline (V , )  P max
i
i
Ramp Rates
Pireg  Ri reg
Pires  Ri res
Bidding into a Bilateral Market
• Objective: Identify suppliers’ Nash Equilibrium
(NE) bidding strategy in a bilateral market. Study
the characteristics of NE bidding strategies.
• Assumptions:
–
–
–
–
m-generator-n-load
Each generator can supply at most one load
Generators submit bids to each load
Each load accepts the cheapest bid generator at its bid price
 Problem: How should a generator set the bid
price for each load?
Optimization to Find NE Prices
• Calculation of NE bid prices for G1:
 3 1.5
5 3 


C 6 2 


10



  12 
Lowest cost of a combination without g:
CC*1 
[c(i,1)  c( j,2)]
i 2 , 3, 4 ,5
j 2 , 3, 4 ,5, j i
 c(2,1)  c(3,2)  7
G1’s profit margin:
G1’s bid to L1:
min
 (1)  CC  CC  7  5  2
*
1
*
b(1,1)  c(1,1)   (1)    5  
Bidding into a Spot Market
• Objective: Formulate electricity spot market from
supplier’s viewpoint. Identify supplier’s optimal bidding
decisions as market conditions change.
Supplier’s information set
Market statistics
Proprietary database
• Spot price
• Load demand
• Load forecast
• Cost curves
• Resource constraints
• Information about competitors
Bidding decision-making
Bid options
• 50MWh @ $20/MWh for peak hours in the next day
• 100MWh @ $18/MWh for an entire day
• ...
 Problem: Which bidding option is optimal for the market status?
Optimal Bidding Decisions
• Markov Decision Process (MDP) to identify
optimal bidding strategy over a planning horizon
N k k
 Value iteration: v (n  1)  max  p [r  v (n )]
i
j
k j  1 ij ij
N k
N k k
k
k
 max [q   p v ( n )] where q   p r
ij j
i
ij ij
k i
j 1
j 1
At state i:
Competitors’
model
Decision option a:
50MW@$26/MWh
Competitor k’s possible bids:
50MW@$23/MWh, prob.= 0.25
70MW@30$/MWh, prob.= 0.3
...
Pija = probability that market moves to state j from state i
rija = profit when market moves to state j from state i
Flexible Contract Pricing
• Objective: Determine the price of a flexible
contract based on stochastic market model.
• Contract parameters:
T
– Contract volume, V (MWh)
( xt  V )
2
t T1
– Starting-time, T1 and ending-time T2
– Maximum power that can be drawn in the t’th time-period: Ct (0  xt  Ct )
– There is a minimum time between time of scheduling decision and
time of actual delivery of energy.
 Problem: How much is this contract worth?
When and how much to deliver?
Optimization to Find Flexible
Contract Pricing
No-arbitrage pricing: Since a buyer can
resell into the spot market, if the buyer
follows the optimal schedule, (s)he expects
to make $800 from the spot market.
Flexible contract price = $800
States at
stage t
States at
stage t+1
1
1
i
i
Optimization : Max. exp. resale revenue
T2
max E (  xt pt )
s.t.
t T1 T
2
 xt
Buyer
V
Pt=$30/MWh
V’=800MWh
t T1
0  xt  Ct
xt : Schedule decision in period t
N
N
Risk Management in a
Competitive Market
• The portfolio together with the
operations in the spot-markets
will give a profit at the end of
a time-horizon.
• Ahead-of-time, the profit is
uncertain, due to fluctuating
prices and demand.
• Decision-makers are riskaverse.
Probability density
of the profit
STD
E
 Problem: What portfolio should a decision-maker
choose?
Hedging through Optimization
Expected
value
Efficient
Frontier
Physical
production
Hedging by
financial instruments
Standard
deviation
Hedging Using Production
• Consider profile of fixed sales at spot market
prices. Can the profile of sales be chosen to
minimize the variance?
• Mapping out efficient frontier:
min
s
sC p , p s T  bC c ,c bT  2 sC p ,c bT
constr.
b  f ( s)
 
 
s p T  b cT 
si  [0, Cap]
Defense Plans in the SPID System
Fast and on-line
power & comm.
System
assessment
Self-healing
Strategies
Power
Infrastructure
Real-Time
Failure
Analysis
Hidden Failure
Monitoring
Robustness
Security
Adaptive load
shedding,
generation
rejection,
islanding,
protection
Vulnerability
Assessment
Dependability
Information
And
Sensing
•Satellite, Internet
•Communication
system monitoring
and control
• Problem: Design self-healing strategies and adaptive
reconfiguration schemes to minimize the impact of
power system vulnerability
Monitoring and Control with a MultiAgent System
Vulnerability
Assessment
Agents
DELIBERATIVE LAYER
Hidden
Failure
Monitoring Agents
Reconfiguration
Agents
Comm.
Agent
Knowledge/Decision
exchange
Restoration
Agents
Event
Identification
Planning
Agent
Agents
Triggering Events
COORDINATION LAYER
Event/Alarm
Filtering
Update Model
Agents
Plans/Decisions
Model
Update
Agents
Check
Consistency
Controls
Events/Alarms
Frequency
Stability
Agents
Fault
Isolation
Agents
Inputs
REACTIVE LAYER
Protection
Agents
Command
Interpretation
Agents
Inhibition Signal
Controls
Power System
Generation
Agents
Control of Available Transfer
Capabilities (ATC)
• ATC definition:
Total Transfer Capability (TTC) Transmission Reliability Margin
(TRM) - Scheduled Capability
ATC
ATC
Scheduled
• FACTS expands TTC and ATC.
Scheduled
TRM
W/O FACTS
• Problem: Increase power transfer capability of
transmission systems using FACTS control
TRM
W/ FACTS
Dynamic Security Based
FACTS Control
• ATC calculation procedure
V
incorporates thermal,
generator and voltage security
constraints.
• EPRI ETMSP simulates
system dynamics.
• Multiple and simultaneous
transfers need to be included.
P
Multi-Agent Coordinated
Control System
System Voltage Control
Agent including OPF
Algorithm
Bus ‘I’ Voltage
Control Agent
Bus ‘J’ Voltage
Control Agent