power market participation of flexible loads and reactive power...
TRANSCRIPT
Power Market Participation of Flexible
Loads and Reactive Power Providers: Real
Power, Reactive Power, and Regulation
Reserve Capacity Pricing at T&D Networks
DIMACS, Rutgers U
January 21, 2013
Michael Caramanis
2
Outline
• How can Flexible Loads Provide Fast Reserves
• How do Distribution Network Injections Differ From Transmission System Bus Injections?
• Current Market Bidding Rules Motivate Flexible Distributed Loads to Exercise Strategic Behavior Resulting in a Hierarchical Game
• Conditions for Hierarchical Game to Converge
• Revised Bidding Rules Remove Strategic Behavior Incentives and Allow ISO/“DNO” to Clear Market in a Socially Optimal Manner
• Detailed Distribution Market Pricing Real and Reactive Power
3
Outline
• How can Flexible Loads Provide Fast Reserves
• How do Distribution Network Injections Differ From Transmission System Bus Injections?
• Current Market Bidding Rules Motivate Flexible Distributed Loads to Exercise Strategic Behavior Resulting in a Hierarchical Game
• Conditions for Hierarchical Game to Converge
• Revised Bidding Rules Remove Strategic Behavior Incentives and Allow ISO/“DNO” to Clear Market in a Socially Optimal Manner
• Detailed Distribution Market Pricing Real and Reactive Power
4
Characteristic Generation Demand
Dispatchability-
Schedulability
Low/Med/High
Wind, Run of Riv /Neuclear,
L.E.P,/ HydroFossil
Capacity Loads, dependent on
env. e.g.,Light/ Ind. Energy
Loads Aluminum
idle/Schedulable production of
electr. energy intensive storable
products (gas liquif.)
Flexibility
Low/med/high
No Ramp – steady output
e.g., nucl, min gen, start up
cost and delay/ Inertia and
medium storage/high ramp-
low inertia large storage
Thermal or work inertia (Allum.
Smelter)/Enegy Demand with
small storage to capacity ratio
(HVAC)/ Large storage to
capacity ratio (ice, molten salt,
batteries in Evs)
Forecastaility
Low/Med/High
Wind, Solar. RoR
Hydro/reliable
fossil/unreliable fossil
Inflexible loads (lighting
cooking)/Weather
dependent/scheduled loads
Voltage Control Synchronous Generators with
dynamic Var compensators,
DC-AC Converters
Distributed Power Electronics
accompanying EVs. HVAC,
Roof top PV.
Generation and Demand Share Functional Characteristics that are
Key to the Efficient and reliable Operation of the Electricity Grid
5
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
D: 10/1/11
S: 9/1/10
Instance of PJM Regulation Signal, y(t). Note Constant Average over relatively
short period of Time
Flexible Loads Require Energy by some deadline =>
Capable of Regulation Reserves
6
Example of Generator providing Super Fast Reserves: Frequency control and 40MW of
Secondary Reserves
Source: Courtesy of EnThes Inc., March 2007
Today Generating Units are Only Reserve Providers
Frequency Control
Secondary Reserves
320MW50MW
7
Outline • How can Flexible Loads Provide Reserves
• How do Distribution Network Injections Differ From Transmission System Bus Injections?
• Current Market Bidding Rules Motivate Flexible Distributed Loads to Exercise Strategic Behavior Resulting in a Hierarchical Game
• Conditions for Hierarchical Game to Converge
• Revised Bidding Rules Remove Strategic Behavior Incentives and Allow ISO/“DNO” to Clear Market in a Socially Optimal Manner
• Detailed Distribution Market Pricing Real and Reactive Power
Distribution Network Low Voltage Bus Marginal Cost Based
Dynamic Prices (DLMP) Result from Augmenting Transmission
System High Voltage Prices (LMP) by Marginal cost of: Line
Losses, Reactive Power, Voltage control, Transformer Life Loss
8
HV, Bus n
LV, n(k) LMP at Bus n
DLMP at n(i)=mn(i)(LMP at n)+…Where mn(i)=(1+ML at n(i))…
LV, n(i)
9
Outline
• How can Flexible Loads Provide Reserves
• How do Distribution Network Injections Differ From Transmission System Bus Injections?
• Current Market Bidding Rules Motivate Flexible Distributed Loads to Exercise Strategic Behavior Resulting in a Hierarchical Game
• Conditions for Hierarchical Game to Converge
• Revised Bidding Rules Remove Strategic Behavior Incentives and Allow ISO/“DNO” to Clear Market in a Socially Optimal Manner
• Detailed Distribution Market Pricing Real and Reactive Power
Examples of Flexible Loads: State
Dynamics Determine Preferences
10
• Distributed PHEV Charging
• Centralized Pumped Storage Hydro Units
1
( ) ( ) ( ) ( )
dep time
( )
ˆ
0
j j j j
j
F F F F
F
t t t t
n i n i n i n i
n i
x x d
x
1
( ) ( ) ( ) ( ) ( )
0 2
( ) ( ) ( )
)
4
( ( )
psh psh t g t r R t
n p n p n p
ps
t t p
n p n p n
h
p n p n
p
p
n p p
h
n
s
x x p g g
x x
11
Strategic Flexible PHEV Load Behavior
, ( )( ) ( )
,
( ) ( ), ,
,, , ,
,
( ) ( ) ( ) ( )
1
( ) ( ) ( ) ( )
( ) ( )
{ [ λ
λ ] ( )}
. .
ˆ e.g., state dyn of EV dem.
min
j j
j
E t R t tF F R n n n it tj j
n i n i
j
j j j j
j j
F F
F F F F
F F R
Ft E t t
n i n n im
j td d j t
F Rt R t t t t
n i n n i n i n i
t t t t
n i n i n i n i
t t
n i n i
E m d
m d U x
s t
x x d
d d
λ λ
,
( ) ( ) ( )
up/dn nature of Reg. Res.
ˆ[ ] Local Constraintj jF F Rt t t
n i n i n i
j
d d C
Use of Current Bidding Rules to
Self Schedule
12
,* *
( ) ( )
, *
( ) ( )
Bid Energy at a very high price
Bid Energy 2 at energy price ~ λ
and Regulation Service Rate at 0.
j j
j
F F Rt t
n i n i
F R t t E t
n i n i n
d d
d m
Using the Current Bidding Rules. Bids described on the previous
slide, induce the ISO/DSO to almost surely Schedule Energy and
Reserves to the * values, and thus effectively self dispatch.
13
( ) ( ) ( )
*
( ) ( ) ( )
( ) ( ), ( )
,
( )
( )
( ) ( ) ( ) ( ) ( ) ( ), , ,
, ,
( ) * 2
( )
, *
( ) ( )
( ) , ( )
. .
0,
max
( )2
( )
j
c t t R tn i n n
j
j
c c R
Ft t c t
n n i n i
n n i j n i
c t E u t
n i n
n i
t t t t t t
n i n i n n n nd g g t
t i
Fn i t
n i
j
F RR t t
n n i
n j n i
s t
g d d
d t
u d c g r g
d
g d
,
( )
( )
& other capacity and ramp constr. for conv. gen. and dem.
R u t
n
c t
n i
n i
tLoss R
14
Outline
• How can Flexible Loads Provide Reserves
• How do Distribution Network Injections Differ From Transmission System Bus Injections?
• Current Market Bidding Rules Motivate Flexible Distributed Loads to Exercise Strategic Behavior Resulting in a Hierarchical Game
• Conditions for Hierarchical Game to Converge
• Revised Bidding Rules Remove Strategic Behavior Incentives and Allow ISO/“DNO” to Clear Market in a Socially Optimal Manner
• Detailed Distribution Market Pricing Real and Reactive Power
Hierarchical Game Dynamics
• Undamped Oscillations when Flex Load
Updates Clearing Price Estimates
Myopically to Most Recent ex-post
ISO/DMO value
• Convergence to stable equilibrium when
Flex Load Updates Clearing Price
Estimates Factoring in History, for
example sets them Equal to their Time
Average 15
16
0.00%
1.00%
2.00%
3.00%
4.00%
5.00%
1 5 91
31
72
1 25 29
33
37 41 45
49
53 57 61
65
69 73 77
81
85 89 93
97
101
105
109
113
117
121
125
129
133
137
141
145
149
153
157
161
165
169
173
% C
on
verg
en
ce
Iteration
UPQB LMP Convergence by Iteration (No Congestion)
Base Case
Discrete Smooth Case
Quadratic Case
17
0.00%
1.00%
2.00%
3.00%
4.00%
5.00%
1 5 913 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97
101
105
109
113
117
121
125
129
133
137
141
145
149
153
157
161
165
169
173
% C
on
verg
en
ce
Iteration
UPQB LMP Convergence by Iteration (With Congestion)
Base Case Bus 1
Discrete Smooth Case Bus 1
Quadratic Case Bus 1
18
0.00%
0.50%
1.00%
1.50%
2.00%
2.50%
3.00%
1 5 913 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97
101
105
109
113
117
121
125
129
133
137
141
145
149
153
157
161
165
169
173
% C
on
verg
en
ce
Iteration
Step Size Impact on UPQB Convergence
Discrete Smooth Case Bus 1 (Steepest Descent)
Discrete Smooth Case Bus 1 (Averaging)
Quadratic Case Bus 1 (Steepest Descent)
Quadratic Case Bus 1 (Averaging)
19
(0.50)
(0.40)
(0.30)
(0.20)
(0.10)
-
0.10
0.20
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
$/M
Wh
Hour
(UPQB LMP - TCCB LMP) by Iteration
Bus 1 Iterations 148, 151, 154, etc
Bus 1 Iterations 149, 152, 155, etc
Bus 1 Iterations 150, 153, 156, etc
20
Outline
• How can Flexible Loads Provide Reserves
• How do Distribution Network Injections Differ From Transmission System Bus Injections?
• Current Market Bidding Rules Motivate Flexible Distributed Loads to Exercise Strategic Behavior Resulting in a Hierarchical Game
• Conditions for Hierarchical Game to Converge
• Revised Bidding Rules Remove Strategic Behavior Incentives and Allow ISO/“DNO” to Clear Market in a Socially Optimal Manner
• Detailed Distribution Market Pricing Real and Reactive Power
Under New Bidding Rule allowing Flex Load to Express True Utility, ISO/DNO will Solve
21
,
( ) ( ) ( ) ( ) ( ) ( )
( )
( ) ( ), , , , , , , , , ,
( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) ( )
( ) ( ), ( )
( )
( )
[
( )] . .
( )2
maxF F Rc t t R t t t W tj j
n i n n n i n i n
j j
j
j
c c
F FR
c t
n i
t t
n i n id g g d d g t i t
t t t t t t
n n n n n i n i
Ft t c t
n n i n i
n n i j n i
Fn i t
n i
j
d
u d
c g r g U x s t
g d d
d
( )
( )
( )
2 ,
( ) ( ) ( )
( ) ( ), ( )
( ) 2 ,
( )
1
( ) ( ) ( ) ( )
0,
( ) 0,2
ˆ ( ) Flex Load Dynamics
& other Local ( ) and System constrai
j
j
j j j j
n i
c t
n i
n i
F F F F
E x t
n
Ft t c t
n n i n i
n n i j n i
Fn i t E x t
n i n
j
t t t t
n i n i n i n i
d
t
g d d
d t
x x d n i
n i
nts
Complex Bid ISO/DNO Market Clearing
Achieves Hierarchical Game Equilibrium
22
• Theorem:
– First order Optimality Conditions
– Complementary Slackness, and
– Feasibility Conditions
Coincide if we combine Hierarchical game
problems and compare to ISO/DSO problem,
Except when Flex loads dominate in a Distr.
Location (competitive assumption fails?)
( )( ) ( )
, , ,
( ) ( ) ( ) () )
( )
,
(
£ 1
0
( )( )j
tn
Fji
j
F E t
ISO nd
R
c t
t t t j t j t
n n i n i n i
t
n i n i n i
j
F R t
n i n i
d d
d
Additional term in ISO/DNO problem
23
( )( ) ( )
, , ,
( ) ( ) ( ) () )
( )
,
(
£ 1
0
( )( )j
tn
Fji
j
F E t
ISO nd
R
c t
t t t j t j t
n n i n i n i
t
n i n i n i
j
F R t
n i n i
d d
d
( )
( ) ( )
,
Becomes negligible, i.e. 0 as
0
or as the relative size of flex load Reg. Res. Transactions 0
j
j
t
n i
t c t
n i n i
F R
F
d
d d
24 (250)
(200)
(150)
(100)
(50)
-
50
100
(0.40)
(0.30)
(0.20)
(0.10)
-
0.10
0.20
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
$/M
Wh
$/M
Wh
Hour
Impact of Competitiveness Assumption
UPQB Price Spread - TCCB Price Spread
UPQB IFL Costs - TCCB IFL Costs
Conclusion • Flexible Loads at Distribution Level may
participate in Expanded ISO/DNO Centrally
Cleared Power Market bringing significant
benefits, particularly w.r.t. Sustainable
Renewable Generation Integration to the Grid
• Expanded ISO/DNO-Operated Power Market
Clearing is Practical from Information and
Computational Tractability Point of view.
• Inclusion of Other Important Distribution Network
Costs, such as Reactive Power Compensation
and Voltage Control is also Practical. 25
26
Outline
• How can Flexible Loads Provide Reserves
• How do Distribution Network Injections Differ From Transmission System Bus Injections?
• Current Market Bidding Rules Motivate Flexible Distributed Loads to Exercise Strategic Behavior Resulting in a Hierarchical Game
• Conditions for Hierarchical Game to Converge
• Revised Bidding Rules Remove Strategic Behavior Incentives and Allow ISO/“DNO” to Clear Market in a Socially Optimal Manner
• Detailed Distribution Market Pricing Real and Reactive Power
Ex. of Var. Speed HVAC - PV Collaboration:
Action in the small by Distr. Flex. Loads
27
Load and Other Resources May Participate fully in
Future Distribution Markets:
28
Centralized Dispatchable Generation Distributed Consumers
Centralized Non-Dispatchable Generation Distributed Generation
e.g., Wind Parks
Centralized Prosumers (e.g., cogeneration) Distributed Prosumers
Current market: Only Centralized generation is a full market participant. All others communicate their
capabilities and needs without feedback and response
Market
Centralized Dispatchable Generation Distributed Consumers
Centralized Non-Dispatchable Generation Distributed Generation
e.g. Wind Parks
Centralized Prosumers(e.g., cogeneration) Distributed Prosumers
Future Market: Many more non-dispatchable cedtralized generators, distributed generators and prosumers.
On the distributed side, “feedback” renders Non-Dispatchable generation and distributed consumers-
producers (prosumers) full market participants
Market
29
Example of a Radial Distribution Network: One Medium Voltage Branch is Shown
with three feeders, each with three building loads. Substation is the Slack Bus
Distribution Market Problem formulation: Minimize Utility Loss, Real and React. Power Cost (incl Losses), Asset Life
Loss, and Volt. Control Cost s.t. Load Flow , Capac., Volt. Magnitude Const
, ,
, ,
22 2
, , (1) , (1) 1m n m ni i i
i M M
m n m n
f fg d dP P P V
g m m m m b b
m i i f f
tMc P u P P C C Q c V
Minimize
Subject to
30
Distribution Market Problem formulation (cont.)
31
,
,
, , ,
,
2
1, 2, , ,
15000 15000exp ,
383 273
,
,
0
b m
b m
b m b m b m
f b mH
f
H A
f f f b m b m
b b b
f
k k S f
V V V b
A
Distribution Market Benefits • Marginal Losses Reflected in DLMPs=>Demand
Adaptation
• Reactive Power Pricing motivates Dual Use of
Power Electronics whose presence is expected to
Become Ubiquitous while accompanying
Distributed Clean Generation (e.g., PV)
installations and Flexible Loads (e.g., EVs, Heat
Pumps)
• Marginal Voltage Control Cost Reflected in DLMPs
=>Demand Adaptation
• Distribution Asset Degradation Marginal Costs
Reflected in DLMPs =>Demand Adaptation
• Full Distr. Net Price Unbundling =>Distr. Net Rent 32