Competitive Demand Response Trading Mechanism using Bi-level Optimization Framework.
Keywords: Demand-side management, electricity pricing, auction-based demand response mechanism design, bi-level optimization, game theory, power system network economics.
Goal: The research project aims to identify the day-ahead operational behavior of market participants both from supply and demand sides of a trans-active market mechanism. We want to provide a coordinated DR exchange (DRX) mechanism, integrated it into a market clearing model to suppress locational marginal price (LMP) spikes and operating cost.
Competitive Demand Response Trading in Electricity Markets: Aggregator and End-User Perspective
1. Competitive Demand Response Trading
in Electricity Markets: Aggregator and
End-User Perspective
NUR MOHAMMAD
BSc (EEE) MSc (EEE)
Presentation for the final seminar of
Doctor of Philosophy
Supervised by
Dr. Yateendra Mishra and Prof. Gerrard Ledwich
School of Electrical Engineering and Computer Science
Queensland University of Technology
December 18, 2017
2. Presentation Outlines
➢ Background, Why Demand Response (DR)………………………………..….…Slide 3,4
➢ Electricity Price on the Rise, How Fluctuating the Price is?................Slide 5,6
➢ Policy and Tariff Reformation………………………………...…………...............Slide 7
➢ Hypothesis, Research Questions, Literature Review, Objectives………Slide 9-11
➢ The Proposed Market Model and Optimization……………………….….….Slide 13-15
-A Framework of DRX Integrated Market Clearing Model (MCM)
-Bi-level Optimization, Solution, Complexity
➢ Model#1, Problem Formulation without Wind Energy Firm (WEF)...Slide 17-20
- Upper-Level (SCED Problem), Lower-Level (DRX Problem)
- Flow Chart, GenCos Revenue and Aggregator’s Payoff
➢ Data, and Simulation Results#1………………………...................................Slide 22-31
- Network, its Data
- Bidding Data of GenCo and Aggregators
- Operation Cost, LMP, Gen Dispatch,
- DR Supply, Payoff for Aggregators and Compensation for Users, Summary
➢ Model#2, Problem Formulation with a WEF………………….……….………Slide 33-34
- Modified Optimization Model
- Wind-DR Pairing, Conditional Priority Scheme
➢ Data, and Simulation Results#2…………………………………………….………….Slide 36-41
- Network, its Data , Case Formation
- Results, Discussion, and Summary
➢ Conclusions, Limitation, and Future Works…………..............................Slide 44-46
2
3. • Demand Response (DR)
• A reduction/shifting of electricity usage from usual consumption at a
certain times to balance/reduce the gap between supply and demand of
electricity.
• DR sees to adjust load demand instead of adjusting generation supply.
• Like pseudo generation from demand side.
• User may reduce electricity consumption in response to (1) price change
over time, (2), financial incentives as compensation. [Dept. of Energy, USA 2005]
Background - Demand Response (DR)?
3
7. Policy and Tariff Reformation
How DR
Service
Provider Get
Rewarded?
DR selling
price should
be at LMP
• In 2010, FERC
order 745
• Practices in
PJM,
• CAISO
Dynamic Price: Real-
Time (RT), ToU
• ComEd’s
(Commonwealth
Edition)
• 5% customers left
the RT in 2014
• Due to RT
hikes/spike
Incentive:
Direct Load
Control (DLC)
• Undesirable
interruption
• Users are less
committed
• Misreporting
• Less visible from
load control
perspective
LMP: Locational Marginal Price
FERC: Federal Electricity Regulatory Commission
7
8. Presentation Outlines
8
➢ Background, Why Demand Response (DR)………………………………..….…Slide 3,4
➢ Electricity Price on the Rise, How Fluctuating the Price is?................Slide 5,6
➢ Policy and Tariff Reformation………………………………...…………...............Slide 7
➢ Hypothesis, Research Questions, Literature Review, Objectives……Slide 9-11
➢ The Proposed Market Model and Optimization……………………….……..Slide 13-15
-A Framework of DRX Integrated MCM
-Bi-level Optimization, Solution, Complexity
➢ Model#1, Problem Formulation without WEF……...............................Slide 17-20
- Upper-Level (SCED Problem), Lower-Level (DRX Problem)
- Flow Chart, GenCos Revenue and Aggregator’s Payoff
➢ Data, and Simulation Results#1………………………..................................Slide 22-31
- Network, its Data
- Bidding Data of GenCo and Aggregators
- Operation Cost, LMP, Gen Dispatch,
- DR Supply, Payoff for Aggregators and Compensation for Users, Summary
➢ Model#2, Problem Formulation with a WEF………………….……………..…Slide 33-34
- Modified Optimization Model
- Wind-DR Pairing, Conditional Priority Scheme
➢ Data, and Simulation Results#2…………………………………………….………….Slide 36-41
- Network, its Data , Case Formation
- Results, Discussion, and Summary
➢ Conclusions, Limitation, and Future Works…………..............................Slide 43-46
9. Hypothesis, Research Questions
Hypothesis
• “A coordinated DR exchange (DRX) mechanism with a
bid based compensation price setting from end-users
can suppress LMP spikes and operating cost.”
Two research questions to be answered
• (1), If a DRX mechanism is integrated then how we set
it’s involvement margin (the amount of DR to be
traded) from network perspective ?
• (2), How can we use DR to smooth wind energy
variability and what its impact? Provided that, a large-
scale wind energy firm (WEF) is in the system.
9
10. social-welfare
Scalability issue
Distribution side, MCM
Profit of wind producer
practical Setting, MCM,LMP
Option DR, Risk, WEF, CVaR
Cost Min, SW Max
silent about DR and WEF
SEUC, DRX, Price-based, LMP
Max Social Welfare
price setting, No TX
DRX concept, UMCP
Cost Min, Revenue of LSE
A single LSE, DR growth
MCM, Coupon-based, LMP
Max SW
Demand-side passive
Network Constraint Impact, LMP
ProsConsObj.ResearchGapProsConsObj. Literature Review
To develop a DRX integrated Market Clearing Model (MCM) for operating
cost and LMP reduction with a goal setting to compensate the end-users
upon receiving the offer from independent DR aggregators
10
11. Research Objectives
1) Providing a bid based DR compensation price and
quantity settlement mechanism upon receiving the DR
offer from the aggregators on behalf of the end-users.
2) Analyzing the operational and economic impacts
(operating cost, emission, and LMP) of demand
flexibility.
3) Providing a detailed DRX framework integrated it into
a security constraint market clearing model (MCM).
4) Quantify how aggregator may get benefitted if the
GenCos exercise its conflicting economic interest to
uplift market price.
5) Analyzing how the DR paired with WEF smooth wind
power variation and what are the cost benefit effects.
11
12. ➢ Background, Why Demand Response (DR)………………………………..….…Slide 3,4
➢ Electricity Price on the Rise, How Fluctuating the Price is?................Slide 5,6
➢ Policy and Tariff Reformation………………………………...…………...............Slide 7
➢ Hypothesis, Research Questions, Literature Review, Objectives………Slide 9-11
➢ The Proposed Market Model and Optimization………………………..…..Slide 13-15
-A Framework of DRX Integrated MCM
-Bi-level Optimization, Solution, Complexity
➢ Model#1, Problem Formulation without WEF…….................................Slide 17-20
- Upper-Level (SCED Problem), Lower-Level (DRX Problem)
- Flow Chart, GenCos Revenue and Aggregator’s Payoff
➢ Data, and Simulation Results#1………………………...................................Slide 22-31
- Network, its Data
- Bidding Data of GenCo and Aggregators
- Operation Cost, LMP, Gen Dispatch,
- DR Supply, Payoff for Aggregators and Compensation for Users, Summary
➢ Model#2, Problem Formulation with a WEF………………….……..…………Slide 33-34
- Modified Optimization Model
- Wind-DR Pairing, Conditional Priority Scheme
➢ Data, and Simulation Results#2…………………………………………….………….Slide 36-41
- Network, its Data , Case Formation
- Results, Discussion, and Summary
➢ Conclusions, Limitation, and Future Works…………..............................Slide 43-46
Presentation Outlines
12
13. A Framework of DRX Integrated MCM
[SCED]
supply
schedule
SCED – Security Constraint Economic Dispatch
DRX – Demand Response Xchange
[DRX]
DR schedule
Upper-Level
Lower-Level
13
14. Bi-Level Optimization
Upper-Level
Problem
solved by EMO
Min Upper-Level Objective Function
Subject to
Upper-Level Constraints
Lower-Level
Problem
Solved by DRXO
Min Lower-Level Objective Function
Subject to
Lower-Level Constraints
[SCED]
[DRX]
SCED – Security Constraint Economic Dispatch
DRX – Demand Response Xchange
• Network security constraints
• Supply-demand constraint
• GenCos supply limit constraints
• DR requirement in buses
[Operation Cost]
OPC
[DR Transaction Cost]
DRTC
• Regulatory constraint
• DR limit per user type
• Aggregator’s payoff limits
14
15. Conversion of Bi-Level Optimization
A single-level
MPEC
Min Upper-Level Objective Function
Subject to
Upper-Level Constraints
KKT Optimality Conditions
associated with Lower-Level Problem(s)
MPEC – Mathematical Program with Equilibrium Constraints
KKT Optimality Conditions:
1). Primal and dual feasibility constraints,
2). Equalities obtained from differentiating the Lagrangian w.r.t. variables,
3). Complementary conditions
KKT – Karush-Kuhn-Tucker
[Dreves et al. 2011; Karush & W 2008]
[Hu & Ralph 2007]
15
16. Presentation Outlines
➢ Background, Why Demand Response (DR)………………………………..….…Slide 3,4
➢ Electricity Price on the Rise, How Fluctuating the Price is?................Slide 5,6
➢ Policy and Tariff Reformation………………………………...…………...............Slide 7
➢ Hypothesis, Research Questions, Literature Review, Objectives………Slide 9-11
➢ The Proposed Market Model and Optimization……………………….……..Slide 13-15
-A Framework of DRX Integrated MCM
-Bi-level Optimization, Solution, Complexity
➢ Model#1, Problem Formulation without WEF……...........................Slide 17-20
- Upper-Level (SCED Problem), Lower-Level (DRX Problem)
- Flow Chart, GenCos Revenue and Aggregator’s Payoff
➢ Data, and Simulation Results#1………………………..................................Slide 22-31
- Network, its Data
- Bidding Data of GenCo and Aggregators
- Operation Cost, LMP, Gen Dispatch,
- DR Supply, Payoff for Aggregators and Compensation for Users, Summary
➢ Model#2, Problem Formulation with a WEF………………….………………..Slide 33-34
- Wind-DR Pairing, Conditional Priority Scheme
- Modified Optimization Model
➢ Data, and Simulation Results#2…………………………………………….………….Slide 36-41
- Network, its Data , Case Formation
- Results, Discussion, and Summary
➢ Conclusions, Limitation, and Future Works…………..............................Slide 43-46
16
17. Upper-Level Problem (SCED Model)
DRTCMinarg0
,0
0
,
},{
,),(B
B)1(
:Subject to
)(Min
1
maxmin
maxS,minS,maxmin
maxmin
1
mk
wik
iii
up
ngg
dn
n
g,kg,knnkn
F
kijijij
ijjkikb
i
kjkikb
n i
iknk
mk
m
d
mk
n
ggn
d
d
,
kRPPR
PgPgPg
FFF
kjiF
λdPg
dλPPc
nknk
nknk
R
b
N
NN
N
b
rg
Transmission line
Generation limit
Ramp rate limits
Demand-supply
balance
constraint
DR constraints
GenCos cost
DR Transaction Cost
[DRTC]
total DR amount
demand at Bus i
The LMP at Bus i
dual variable for
line congestion charge
dual variable for
gen. capacity limits
i: index for bus
n: index for supply
k: index for time
17
Minimizing cost
18. Lower-Level Problem (DRX Model)
max
maxmin
),|(
),(
0
,
,A
AA
:Subject to
)(DRTCMin
m
k
m
1-k
mj
k
mj
mj
k
mjmj
d
jmik
jmm
k
mja
k
jmr
k
mja
k
mj
jmm|
k
mjum
dd
.ddd|
rdr
dχ
dds
|
d
dd
,
a
a
N
N
DR offer price
DR to be transected
buyer design matrix
DR capacity limit
j: index for buyer
m: index for seller
k: index for time
AAA,
10
01
10
01
10
01
A,
100
100
010
010
001
001
A
TT
rara
DR tuning parameter
directed graph
L:= (N, A),
seller design matrix
compensation price
pick-up/drop-off rate
18
19. The Flow Chart
Min Objective DRTC
Subject to
KKT set for Aggregator j1
Min Objective OPC
Subject to
Upper-Level Problem:
• argmin: Lower-Level Problem [DRX]
Lower-Level Problem:
• Upper-Level Constraints [SCED]
• Upper-Level Constraint
KKT set for Aggregator j2
KKT set for Aggregator jNa
• Stationary conditions
• Complimentary slackness
• Primal feasibility
• Dual feasibility
DRTransactionCost
TransactedDR
✓ DR sale prices
✓ DR quantities share
✓ User’s compensation
✓ LMPs
✓ Generation dispatch
✓ Gen’s Profit
19
20. • GenCos Payoff
• Aggregator’s Payoff
tt k
nknkn
k
nknkn PgPgcPg
TT
***
λ )(
LMP at Bus where generation located
compensation
tT Nn T k i
k
jm
d
jm
k
k
jmikA
ut
R **
dλdλ
generation amount settled
generation cost
LMP at Bus i where the DR capable load
sum over DR periods
sum of all users
Payoff of GenCos and Aggregator
revenue generation cost
revenue compensation cost
20
21. Presentation Outlines
➢ Background, Why Demand Response (DR)………………………………..…..…Slide 3,4
➢ Electricity Price on the Rise, How Fluctuating the Price is?................Slide 5,6
➢ Policy and Tariff Reformation………………………………...…………................Slide 7
➢ Hypothesis, Research Questions, Literature Review, Objectives…….…Slide 9-11
➢ The Proposed Market Model and Optimization……………………….……...Slide 13-15
-A Framework of DRX Integrated MCM
-Bi-level Optimization, Solution, Complexity
➢ Model#1, Problem Formulation without WEF……...............................Slide 17-20
- Upper-Level (SCED Problem), Lower-Level (DRX Problem)
- Flow Chart, GenCos Revenue and Aggregator’s Payoff
➢ Data, and Simulation Results#1………………………................................Slide 22-31
- Network, its Data
- Bidding Data of GenCo and Aggregators
- Operation Cost, LMP, Gen Dispatch,
- DR Supply, Payoff for Aggregators and Compensation for Users, Summary
➢ Model#2, Problem Formulation with a WEF………………….………………….Slide 33-34
- Wind-DR Pairing, Conditional Priority Scheme
- Modified Optimization Model
➢ Data, and Simulation Results#2…………………………………………….………….Slide 36-41
- Network, its Data , Case Formation
- Results, Discussion, and Summary
➢ Conclusions, Limitation, and Future Works…………...............................Slide 43-46
21
22. Network Data Used
Table: Generation Capacity limits, Emission rate,
Lines and Load Data [Walawalkar et al. 2008]
Gen
(Fuel type)
Gen Limits
[Pn
min Pn
max ]
(p.u.)
Carbon Emissions
Rate
(kgCO2/kWh)
G1 (Coal), Bus#1 [0.25, 1.10] 0.940
G2 (Diesel), Bus#1 [0.25, 1.00] 0.778
G3 (Coal), Bus#3 [1.50, 5.20] 0.940
G4 (Gas), Bus#4 [0.50, 3.00] 0.581
G5 (Coal), Bus#5 [2.25, 6.00] 0.940
Transmission Lines and Load Data
From
Bus i
To
Bus j
Reacta
nce
(xij)
Capacity
Limit
[Fij
min Fij
max]
Load
Demand,
Di, (p.u.)
1 2 2.81 [-8.75, 8.75] Bus#2=3.00
2 3 1.08 [-8.75, 8.75] Bus#3=3.00
4 3 2.97 [-8.75, 8.75] Bus#4=3.00
5 4 2.97 [-2.40, 2.40]
5 1 0.64 [-8.75, 8.75]
1 4 3.04 [-8.75, 8.75]
Figure: PJM 5-Bus System.
[Li & Rui 2007]
supply
prevails
demand
prevails
22
23. Supply and Demand-Side Bidding Data
Table A: Cost coefficient, bidding quantities and prices for GenCos
Gen Generation supply offer price PJM
[an, bn]
($/MWh, $/MWh2)
[1st Block]
($/MWh),
(p.u)
[2nd Block]
($/MWh),
(p.u)
[3st Block]
($/MWh),
(p.u)
[4th Block]
($/MWh),
(p.u)
[5th Block]
($/MWh),
(p.u)
($/MWh)
G1 [0.115, 11.50] 11.61, 0.5 17.36, 1.1 23.11 28.86 34.61 14.00
G2 [0.115, 11.50] 11.61, 0.5 17.36, 1.0 23.11 28.86 34.61 15.00
G3 [0.355, 12.50] 12.85, 1.0 30.60, 2.0 48.35, 3.0 66.10, 4.0 83.85, 5.0 30.00
G4 [0.425, 18.50] 18.92, 0.6 40.17, 1.2 61.42, 1.8 82.67, 2.4 103.92, 3.0 40.00
G5 [0.265, 10.50] 10.76, 1.2 24.01, 2.4 37.26,3.6 50.51, 4.8 63.76, 6.0 10.00
Aggregator Capacity DR cost coefficients
The DR offer price segments
for the user group at ωm=1
dmax αm βm U1 U2 U3
A1 1.90 0485 12.15 16.97 23.06 27.42
A2 1.70 0.092 10.35 21.71 24.72 26.43
A3 1.70 0.068 11.05 19.20 24.65 24.66
The bidding parameter, ωm in DR price is varied from 1:5.50 with different increment to
rescaled price coefficient, αm and βm at different DR levels.
Table B: The DR bidding prices. ))Γ(1( 2
mjummjmmum dds ,
23
DRTC function
24. Operating Cost with and without DRX
Table: At different level of DR, the operation cost with and without DR, emission.
DR Amount
System
demand
100 Base
% DR
Operating
Cost
Relative
Cost
Reduction
DRX
Transaction
Cost
Emissions
p.u. p.u. % k$ % k$ [tonCO2]
0 202.26 0 747.49 0 0 19077
4.02 198.18 1.95 724.16 3.12 7.75 18806
12.21 194.08 5.92 686.31 8.18 29.34 18167
17.03 189.25 8.26 672.61 10.01 48.76 17764
21.86 184.42 10.60 679.22 9.13 74.32 17335
26.69 179.59 12.94 687.10 8.07 107.28 16881
31.52 174.76 15.28 734.44 1.74 178.48 16427
36.35 169.94 17.72 802.55 -7.36 266.92 15974
39.96 166.32 19.37 909.64 -21.69 382.34 15634
43.57 162.71 21.21 1040.62 39.21 516.08 15294
24
25. Operation Cost Trajectory and LMPs
Table: The average LMP for
different incremental DR.Figure: The operation cost trajectory with
and without DR transaction.
Average LMP ($/MWh)
% DR Bus#1 Bus#2 Bus#3 Bus#4 Bus#5
0 50.51 51.82 55.25 56.49 16.28
1.95 47.92 49.14 52.3 53.46 16.28
5.92 46.3 47.45 50.46 51.55 16.28
8.26 45.76 46.89 49.85 50.92 16.28
10.6 44.14 45.21 48 49.01 16.28
12.94 44.14 45.21 48 49.01 16.28
15.28 44.14 45.21 48 49.01 16.28
17.72 44.14 45.21 48 49.01 16.28
19.37 44.14 45.21 48 49.01 16.28
21.21 44.14 45.21 48 49.01 16.28
25
26. Hourly LMPs over a Day
Figure A: The hourly LMPs at
Bus#3.
Figure B: The hourly LMPs at
Bus#4.
26
27. Generation Dispatch for Three DR Levels
Figure A: Generation dispatch
mix [no elasticity to the demand]
Figure B: Generation dispatch mix
[with 1.95% DR participation].
Figure C: Generation dispatch mix
[with 5.92% DR participation].
27
28. Aggregators’ Payoff Share
Figure: Payoff trajectory of the
aggregators across the incremental DR
levels.
Table: The payoff receive
by each aggregator.
-29
-19
-9
1
11
21
31
41
1.95 5.92 8.26 10.6 12.94 15.28 17.72 19.37 21.21
Aggregator'sPayoff(k$)
% DR of demand w/o DR
J1
J2
J3
Poly. (J1)
Poly. (J2)
Poly. (J3)
Payoff (k$)
J1 J2 J3
15.15 13.75 14.41
23.05 19.35 20.59
26.09 17.33 19.27
22.93 15.89 16.19
26.37 15.98 16.09
2.92 -1.65 -1.47
% DR
1.95
5.92
8.26
10.60
12.94
15.28
+𝑅 𝐴 −𝑅 𝐴
28
29. DR Offer Price and Compensation
Figure A: The DR supply offer
price which increases if
the DR demand rises. 16.97
21.71
19.2
23.06
24.72 24.65
27.42 26.43
24.66
0
5
10
15
20
25
30
35
A1 A2 A3
DRofferprice($/MWh)
U1 U2 U3
0
2
4
6
8
10
12
U1 U2 U3 U1 U2 U3 U1 U2 U3
A1 A2 A3
DR(p.u.)
DR provided by end-user group under each Aggregatror j
% DR of demand w/o DR
1.95 5.92 8.26
10.6 12.94 15.28
17.72 19.37 21.21
Figure B: The DR provided by
each user types at
different incremental
DR levels.
A1 A2 A3
DR U1 U2 U3 U1 U2 U3 U1 U2 U3
1.95 2.31 0 0 2.88 0 0 2.553 0 0
5.92 9.06 0 0 10.76 0 0 9.519 0 0
8.26 17.05 0 0 16.82 0 0 14.88 0 0
10.6 30.14 4.28 0 25.37 1.23 0 23.22 1.22 0
12.94 36.84 14.75 0 30.82 6.59 0 30.73 6.57 0
15.28 52.66 32.97 0 42.37 17.22 0 42.25 17.17 0
Table: DR compensation benefit share
among the user groups
29
30. Impact on Aggregator’s Payoff
All GenCos
Competitive
The G3
Strategic
(Scenario#1)
The G4
Strategic
(Scenario#2)
The G3, G4
Both Strategic
(Scenario#3)
Price Rigging
($/MWh)
$83.80
/$66.10
$103.92
/$82.67
$83.80
$103.92
Aggregator’s Payoff
(k$) without DR
0 0 0 0
Aggregator’s Payoff
(k$) with 5.92% DR
{23.05, 19.35,
20.59}
{23.42,19.70,
20.95,
{21.69, 18.06,
19.31}
{24.72, 20.94,
22.18}
Aggregated Payoff
(k$)
62.29 64.07 63.72 67.84
Relative payoff
variation
(k$) and %
(64.07 - 62.59)
= 1.78 (2.85%)
(63.72 - 62.29)
= 1.13 (1.82%)
(67.84 - 62.29)
= 5.55 (8.90%)
Table: Comparison of aggregator’s payoff due degree of
strategy adopted by GenCos.
NB: Strategic Bidding Hours: 7 pm, 8 pm, and 9 pm
30
31. Summary of Model#1
✓ In Model#1, a DR integrated MCP settled in two levels is modelled.
✓ DRXO determines optimal DR amount traded and the DR
compensation price in lower-level.
✓ The EMO considers the lower-level transacted DR and its cost into
MCM at upper-level to find generation dispatch and the LMPs.
✓ Simulation results compare the operating cost, LMPs, aggregator’s
payoff, the of DR should be transacted, it's cost.
✓ DR compensation benefit consist of the DR compensation price
with the allocated DR is regarded as compensation benefit for the
users
✓ Beyond a critical DR level, the DR compensation price become
higher and outweigh the DR benefit both the upper and lower level
31
32. Presentation Outlines
32
➢ Background, Why Demand Response (DR)………………………………..…..…Slide 3,4
➢ Electricity Price on the Rise, How Fluctuating the Price is?................Slide 5,6
➢ Policy and Tariff Reformation………………………………...…………................Slide 7
➢ Hypothesis, Research Questions, Literature Review, Objectives…….…Slide 9-11
➢ The Proposed Market Model and Optimization……………………….……...Slide 13-15
-A Framework of DRX Integrated MCM
-Bi-level Optimization, Solution, Complexity
➢ Model#1, Problem Formulation without WEF……...............................Slide 17-20
- Upper-Level (SCED Problem), Lower-Level (DRX Problem)
- Flow Chart, GenCos Revenue and Aggregator’s Payoff
➢ Data, and Simulation Results#1………………………................................Slide 22-31
- Network, its Data
- Bidding Data of GenCo and Aggregators
- Operation Cost, LMP, Gen Dispatch,
- DR Supply, Payoff for Aggregators and Compensation for Users, Summary
➢ Model#2, Problem Formulation with a WEF………………….……………….Slide 33-34
- Wind-DR Pairing, Conditional Priority Scheme
- Modified Optimization Model
➢ Data, and Simulation Results#2…………………………………………….………….Slide 36-41
- Network, its Data , Case Formation
- Results, Discussion, and Summary
➢ Conclusions, Limitation, and Future Works…………...............................Slide 43-46
33. SCED Problem Considering Wind
DRTCMinarg0,,0
0
,
,
0
},{
,),(B
B)1(
:Subject to
)()(Min
1
maxmin
max
maxS,minS,maxmin
maxmin
1
1
mkwik
iii
up
nww
dn
n
up
ngg
dn
n
ww
g,kg,knnkn
F
kijijij
ijjkikb
i
kjkikb
n i
ikw
n
wnk
mk
m
d
mk
n
wwn
n
ggn
dD
,
kRPPR
kRPPR
PP
PgPgPg
FFF
kjiF
λdPPg
dλPPcPPc
nknk
nknk
nknk
nk
nknknknk
R,
b
N NN
NNN
N
g bw
rwg
WEF’s offer price
wind power
variable
wind power
availability
limits
WIDR parameter
33
35. Presentation Outlines
35
➢ Background, Why Demand Response (DR)………………………………..…..…Slide 3,4
➢ Electricity Price on the Rise, How Fluctuating the Price is?................Slide 5,6
➢ Policy and Tariff Reformation………………………………...…………................Slide 7
➢ Hypothesis, Research Questions, Literature Review, Objectives…….…Slide 9-11
➢ The Proposed Market Model and Optimization……………………….……...Slide 13-15
-A Framework of DRX Integrated MCM
-Bi-level Optimization, Solution, Complexity
➢ Model#1, Problem Formulation without WEF……...............................Slide 17-20
- Upper-Level (SCED Problem), Lower-Level (DRX Problem)
- Flow Chart, GenCos Revenue and Aggregator’s Payoff
➢ Data, and Simulation Results#1………………………................................Slide 22-31
- Network, its Data
- Bidding Data of GenCo and Aggregators
- Operation Cost, LMP, Gen Dispatch,
- DR Supply, Payoff for Aggregators and Compensation for Users, Summary
➢ Model#2, Problem Formulation with a WEF………………….……………….Slide 33-34
- Wind-DR Pairing, Conditional Priority Scheme
- Modified Optimization Model
➢ Data, and Simulation Results#2…………………………………………………….Slide 36-41
- Network, its Data , Case Formation
- Results, Discussion, and Summary
➢ Conclusions, Limitation, and Future Works…………...............................Slide 43-46
36. Network and Data Used
Table: Generation, Emission, Lines, and Load Data
Figure: Sample 4-bus System
Gen
(Fuel type)
Capacity
[Pn
min Pn
max ]
(p.u.)
Carbon Emissions Rate
(kgCO2/kWh)
G1 (Coal) [3.00 16.50] 0.940
G2 (Gas) [1.50 9.50] 0.581
WEF (Wind) [0.00 4.75] 0.009
Transmission Lines and Load Data
From
Bus i
To
Bus j
Reactance
(xij)
Capacity
[Fij
min
Fij
max]
Load and its
uncertainty (p.u.)
1 4 0.146 [-2.25 3.50] Bus#2=2.00
1 2 0.120 [-3.50 5.75] Bus#3=3.00
2 3 0.120 [-3.50 5.75] Base,
N (1.75,0.12)
4 3
0.100 [-4.00 6.50] Shoulder,
N (3.25, 0.25)
1 3 0.126 [-3.70 5.85] Peak,
N (6.25,0.65)
36
37. Modified IEEE 24-Bus system
[C. Grigg and P. Wong, 1999]
Gen
• The WEF at Bus#9;
Load
• at Bus#1, #3, #6, #7, #13, #15
and #18 are assumed to be
elastic and have DR capability
Line
• interconnectors between Bus#4
to Bus#5 and Bus#5 to Bus#7
System consists of 24 Buses, 38
lines, 17 loads and 11 Gen units.
37
38. Case Formation
• The WEF is modelled as Weibull distribution with scale (c=6.34) and
shape (s=5.68) parameters. [Donk et al. 2005]
• The cut-in (vco), rated (vr) and cut-out (vco) speed are assumed to be
2.25 m/s, 6.75 m/s and 18.15 m/s respectively.
Case#1
(Low-Wind,
High-DR)
v < vL,
vL =2vci
Case#3
(Med-Wind,
Med-DR)
vL < v <vM,
vM = vr
Case#3
(High-Wind,
Low-DR)
v > vH,
vH =1/2vco
38
39. Cases DR Portfolio WEF Portfolio
DR
[%]
Profit
[$K]
Energy
[%]
Profit
[$M]
Emissions
[tonCO2]
Case#1 14.17 35.89 4.150 3.459 28555
Case#2 10.87 26.82 30.11 27.48 22762
Case#3 8.94 16.38 48.80 43.20 17763
Results: Portfolio of WEF and DR
Cases DR Portfolio WEF Portfolio
DR
[%]
Profit,
[$M]
Energy
[%]
Profit
[$M]
Emissions
[M tonCO2]
Case#1 13.98 0.739 9.850 66.73 0.2889
Case#2 11.35 0.680 11.77 92.43 0.2994
Case#3 7.77 0.569 13.54 100.84 0.3011
Table B: Profit of WEF and DR in 24-Bus System.
Table A: Profit of WEF and DR in 4-Bus System.
39
40. Results: User’s Price Rigging Strategy
Groups Benefit [$] Benefit [$] Benefit [$] Benefit [$]
U1 6677.64 6202.92 7760.21 9257.96
U2 5945.64 7026.12 4899.93 8525.82
U3 2763.76 2763.76 4896.31 4126.47
Table B: User’s Compensation Benefit due to
different type value [Case#1 ]
Cases Case#1 Case#2 Case#3
Payoff [$] 9875.72 6644.29 5506.66
DR [U1,U2, U3] 6.5, 5.2, 2.3 6.2, 3.3, 1.0 5.8, 2.6, 0.7
Table A: Aggregator’s Payoff for Different Cases.
%ˆ 0 u %ˆ 101 %ˆ 102 %ˆ 103
40
41. Summary of Model#2
✓ The Model#2 presents how to smooth various wind power levels
using the DR.
✓ A conditional priority scheme is used to improve shortfall and to
minimize operation cost.
✓ The least cost operation is achieved either by increasing DR when
the wind is low or decreasing the DR during high winds.
✓ Simulation results shows the profit of WEF and the aggregators.
The DR, emission, and user’s compensation are compared.
✓ The effect of user’s reported type value on compensation benefit
is investigated.
✓ Due to strategy-proof, no end-users can achieve a higher benefit by
reporting a type value different from its true type value.
✓ Thereby, the end-users get a fair allocation of DR and the
compensation benefits.
41
42. Presentation Outlines
42
➢ Background, Why Demand Response (DR)………………………………..…..…Slide 3,4
➢ Electricity Price on the Rise, How Fluctuating the Price is?................Slide 5,6
➢ Policy and Tariff Reformation………………………………...…………................Slide 7
➢ Hypothesis, Research Questions, Literature Review, Objectives…….…Slide 9-11
➢ The Proposed Market Model and Optimization……………………….……...Slide 13-15
-A Framework of DRX Integrated MCM
-Bi-level Optimization, Solution, Complexity
➢ Model#1, Problem Formulation without WEF……...............................Slide 17-20
- Upper-Level (SCED Problem), Lower-Level (DRX Problem)
- Flow Chart, GenCos Revenue and Aggregator’s Payoff
➢ Data, and Simulation Results#1………………………..................................Slide 22-31
- Network, its Data
- Bidding Data of GenCo and Aggregators
- Operation Cost, LMP, Gen Dispatch,
- DR Supply, Payoff for Aggregators and Compensation for Users, Summary
➢ Model#2, Problem Formulation with a WEF………………….……………….Slide 33-34
- Wind-DR Pairing, Conditional Priority Scheme
- Modified Optimization Model
➢ Data, and Simulation Results#2…………………………………………….………….Slide 36-41
- Network, its Data , Case Formation
- Results, Discussion, and Summary
➢ Conclusions, Limitation, and Future Works…………............................Slide 43-46
43. Conclusions
✓ We studied how to trade DR among multiples aggregators and end-
users in DRX market environment.
✓ A DRX integrated MCM without and with WEF using a bi-level
optimization method is proposed.
✓ The transaction cost, compensation price, and the amount of DR
allocated between the end-users and the aggregator are obtained.
✓ Plausible bidding behavior of the GenCos and its effect on MCM, and
DRX market have been investigated.
✓ DR share causes the reductions in purchases of electricity from the
expensive generators.
✓ When the GenCos bid strategically, aggregator payoff could be higher
than the competitive bidding.
✓ The DR would be profitable as long as its transaction is cost-effective
and economic.
43
44. Conclusions
✓ The proposed WIDR allow WEF to better compete with other
generation company.
✓ WEF and DR pairing pushes operating cost and LMP lower.
✓ Due to strategy-proof, the user may not get more compensation
benefit by misreporting of the type.
✓ Due to receive compensation (payback), the users reduce the energy
consumption cost.
✓ All benefit can be comprehended, if the operating cost reduction at
upper-level does not outweigh the compensation given to the end-
users at lower-level.
***Besides the presentation contents, we have modelled additional two
market mechanism.
✓ (1) Retailer’s Risk-Aware Trading Framework with the DR aggregators.
✓ (2) A DR integrated bi-level transactive market clearing model (TMC).
44
45. Publications
• Conference Papers
(1). N. Mohammad and Y. Mishra, "Competition driven bi-level supply offer strategies in day-
ahead electricity market," 2016 Australasian Universities Power Engineering Conf. (AUPEC),
Brisbane, Australia, 2016, pp. 1-6.
(2). N. Mohammad and Y. Mishra, " Transactive Market Clearing Model with Coordinated
Integration of Large Scale Solar PV Firm and Demand Response Capable Loads," 2017
Australasian Universities Power Engineering Conf. (AUPEC), Victoria, Australia, 2017, pp. 1-6.
• Book Chapter
N. Mohammad, Y. Mishra, “Demand-side Management and Demand Response for Smart
Grid.” In: Kabalci Yasin, Kabalci Ersan, editors. Handbook of Smart Grid Communication
Systems. 1st ed. New York: Springer; will be appear in March 2018.
• Draft Articles to be Submitted
(1) N. Mohammad, Y. Mishra, “A Critical Review on Market Based Demand-Side
Management – PJM and NEM Case Studies.”
(2) N. Mohammad, Y. Mishra, “Incremental DR, GenCos Strategic Bidding and its Impact.”
(3) N. Mohammad, Y. Mishra, “Wind-Induced Demand Response Pairing to Minimize
Operation Cost in Day-Ahead Electricity Markets.”
(4) N. Mohammad, Y. Mishra, “Retailer’s Risk-Aware Trading Framework with DR
Aggregators in Short-Term Electricity Markets.”
45
46. • This work focused on DR trading in transmission and wholesale level.
• Collecting the DR from large-customers is a vast area of research and
could be investigated in future works.
• Especially industrial customers require excessive energy
consumption with normal loads of hundreds of MWs.
• Compare to the residential users, industrial DR would be complex
due to interdependent industrial tasks being difficult to isolate.
• (1) Considering task/process scheduling constraints, DR can be
challenging and need to be studied.
• (2) Further, a transactive control for industrial HVAC cold-storage
may be modelled to participate in electricity markets.
• A small changes in temperature setting have insignificant effect in
plants while it may save huge energy and make a lot of money.
• Adding additional constraints, our proposed model can be modified
to optimize those two models.
Limitation and Future Works
46