Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Takayuki Ishizaki (Tokyo Institute of Technology)
2nd ESIC-HARPS Workshop on Smart Grid and
Renewable Energy
Power Profile Markets: Power Profile Balancing under
Large-Scale Penetration of Battery Storage
Masakazu Koike, Asami Ueda, Tu Bo, Jun-ichi Imura
Introduction Large-scale penetration of battery storage
• improve dispatchability of renewables
Consumer Producer
Agg 1 Agg 2
ISO
time
po
we
r
time
po
we
r
• power profile control is necessary to regulate unpredicted battery use
balance
Power profile market for competitive aggregators
profile = vector
2/18
Ways to Market Clearing
Market mechanism = Math program:
Present: time point market clearing
13:00 13:30 12:00 12:30 14:00 11:30
Prospect: time period market clearing
13:00 13:30 12:00 12:30 14:00 11:30
for efficient
use of battery
Goal: Clearing scheme for non-strictly convex power profile markets
- applicable to vector-valued products
/ premise of strictly convex problem
- simple market clearing by bid funs
/ valid only for scalar-valued products
ISO
Agg 1 Agg 2
Bidding strategies: offline programs Dynamical pricing: online iterations
ISO
Agg 1 Agg 2
Contents
Introduction of Market Mechanisms
Difficulty of Power Profile Market Clearing
Clearing Scheme Based on Particular Property of Electricity Markets
Numerical Verification
Concluding Remarks
4/18
Math Tools from Convex Analysis
Subdifferential: map to set-values if F is not smooth
monotone increasing
Conjugate transformation:
conjugate
inverse
jump
jump
not strictly
convex
not smooth
strictly convex
smooth
&
5/18
Links to Market Mechanisms Profit function:
income cost time
13:00 13:30 12:00 12:30 14:00
focus on a time point
both and are scalars
Maximum profit: Maximizer:
bid function
Lagrangian
consumer
producer
6/18
Bidding Strategy for Profile Markets? Profit function:
income cost time
13:00 13:30 12:00 12:30 14:00
focus on a time period
both and are vectors (profiles)
Bidding strategy is practical only for scalar-valued products
Market clearing with bid functions:
Bid function:
Too hard to plot!!
aggregator α
7/18
Contents
Introduction of Market Mechanisms
Difficulty of Power Profile Market Clearing
Clearing Scheme Based on Particular Property of Electricity Markets
Numerical Verification
Concluding Remarks
8/18
Cost Function of Power Profiles Power profile of Aggregator :
: generation profile
: load profile (constant)
: battery charge & discharge profiles
profiles in
【Lemma】 is convex.
uncertain renewables can be involved as robust/stochastic optimization
: constraints
9/18
Generation cost (convex): Battery usage cost (convex):
including evaluation of final SOC:
Conjecture from Numerical Simulations
Market clearing:
Consumer:
Producer:
Large-scale battery penetration leads to price leveling-off?
Lagrangian
Clearing price when varying the degree of battery penetration:
battery penetration small large
Pric
e
time label
10/18
Mathematical Deduction
【Definition】 A power profile is said to be shiftable between
Price levelling-off deduced from battery capacity margin
【Theorem】 There exists some aggregator such that
is shiftable between time points i and j iff
time points i and j if
po
we
r
shift
shift
time
shiftability without
changing cost
battery!
non-strict convexity
11/18
Pricing via Energy Bid Functions
price
levelling-off
Power profile profit function:
Energy profit function:
Levelling-off price can be found via offline programs
plottable bid function
for energy trade!
Profit maximum energy:
energy consumer
energy producer
bid functions = offline programs
12/18
From Energy Balance to Profile Balance
time
po
we
r
time
po
we
r
profile shift
but and
Energy profit function is maximized
Profile imbalance minimization:
where
【Theorem】 if In particular
Decentralized scheme by ADMM: ISO
Agg 1 Agg 2 imbalance
Contents
Introduction of Market Mechanisms
Difficulty of Power Profile Market Clearing
Clearing Scheme Based on Particular Property of Electricity Markets
Numerical Verification
Concluding Remarks
14/18
Agg 3 Agg 2 Agg 1
Numerical Verification
time label (time length n = 12)
Small Middle Large
pric
e
Consumers
(load + battery)
Producer
(generator)
ISO
scale is not yet adjusted to real situation
Clearing price when varying the degree of battery penetration:
price levelling-off
15/18
Pricing by Energy Bid Functions
price
Small Middle Large
Bid function 3 Bid function 1
Total bid functions:
Bid function 2
tradi
ng a
mou
nt
tradi
ng a
mou
nt
Bid functions in the large battery case:
16/18
discharge
charge
price sensitive
Profile Imbalance Minimization
Agg 3
Aggs 1 & 2
imbalance
time label
Profile imbalance minimization by ADMM:
imba
lanc
e [%
]
iteration step
pow
er
Resultant power profiles:
Small Middle Large
Agg 3
Aggs 1 & 2
imbalance
Agg 3
Aggs 1 & 2
imbalance
17/18
Concluding Remarks
Thank you for your attention!
Power profile markets for competitive aggregators
large-scale battery penetration leads to price levelling-off
profile shiftability owing to battery capacity margin
non-strictly convex programs
levelling-off price is found by energy bid functions (offline programs)
decentralized profile imbalance minimization by ADMM
Future works
analysis on the best length of time periods
Present: time point market clearing
13:00 13:30 12:00 12:30 14:00 11:30
Prospect: time period market clearing
13:00 13:30 12:00 12:30 14:00 11:30
for efficient
use of battery