strategic use of storage: the impact of carbon policy ... · ff by market power, carbon policy,...
TRANSCRIPT
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Strategic Use of Storage: The Impact of
Carbon Policy, Resource Availability, and
Technology Efficiency on a
Renewable-Thermal Power System
Sebastien Debiaa Pierre-Olivier Pineaua Afzal S. Siddiquib
aHEC Montreal
bUniversity College London, Stockholm University, HEC Montreal, and Aalto University,
e-mail address: [email protected]
15 February 2019
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Table of Contents
1 Introduction
2 Mathematical Formulation
3 Comparative Statics
4 Numerical Examples
5 Conclusions
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Introduction
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Motivation
Is Energy Storage the Answer?
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Motivation
Role of Flexible Generation
Marginal costs: e20-22/MWh (lignite-coal), e30/MWh
(CCGT), e45/MWh (gas)
Daily prices: e24.1/MWh and e43.5/MWh
*Residual load = buy volume - wind - solar
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Motivation
“Texas Utility Oncor Wants to Invest $5.2B in Storage: Can It
Get Approval?,” Greentech Media, 10 November 2014
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Related Work
Equilibrium Analysis of Storage
Crampes and Moreaux (2001) show theoretically how market
power may lower water value and even render it negative
Bushnell (2003) finds empirical support via a complementarity
analysis of the California hydro-thermal system
Restructuring and sustainable transition enhance hydropower
producers’ incentives to shift generation between peak and
off-peak periods (Mathiesen et al., 2013)
Sioshansi (2010) analyses the welfare impacts of storage use
via a stylised partial-equilibrium model with perfectly
competitive electricity generation
Sioshansi (2014) demonstrates when storage can reduce social
welfare
In the context of transmission capacity, Downward et al.
(2010) examine how a carbon tax may actually increase
emissions in the presence of market power
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Complementarity 101
Evolving Paradigms
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Complementarity 101
Playing Games
Optimisation Problem constrained by
Optimisation Problems (OPcOP)
Equilibrium Constraints
Mathematical Program with Equilibrium
Constraints (MPEC)
Equilibrium Constraints
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Contribution
Research Objective and Findings
How are storage operations, producer profit, and social welfare
affected by market power, carbon policy, and device efficiency?
Bushnell (2003)’s main findings about more off-peak RE
production and a negative marginal value of RE storage undermarket power hold for only a low level of the carbon tax
Moreover, a carbon tax may actually increase peak thermal
production under perfect competition
Greater storage efficiency monotonically decreases marginalvalue of RE storage under market power but has anon-monotonic effect under perfect competition
RE producer profits from a ceteris paribus increase in the
efficiency of an inefficient device but a “turning point” when the
device is so efficient that the peak price is actually depressed
Social welfare increases monotonically with device efficiency
Conflict between private and social incentives can be partially
resolved through a carbon tax
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Mathematical Formulation
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Assumptions
Assumptions
Two periods, j = 1, 2, that correspond to off- and on-peak,
respectively
No uncertainty or transmission constraints
Equilibrium model with a (i) storage-enabled renewable-energy
(RE) and (ii) thermal producer
Pj(xj + yj) = Aj − (xj + yj) [in $/MWh], where
A2 > A1 > 0
RE production incurs no explicit cost, but available stock of
RE capacity is D ≥ 0 [in MWh] and x1 + Fx2 = D, where
F ≥ 1
Thermal cost function 12Cy2
j [in $/MWh], where C > 0 [in
$/MWh2] with emission rate R > 0 [in t/MWh] and carbon
tax T [in $/t]
For interior solutions, require (i) A2 > FA1 and (ii)
C > 3 − 2√2 ≈ 0.17
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Problem Formulation and KKT Conditions
RE Producer
RE producer’s problem is to maximise its profit over the twoperiods consisting of revenue from sales subject to its resourceavailability
Maximisexj≥0 [A1 − (x1 + y1)]x1 + [A2 − (x2 + y2)]x2(1)
s.t. x1 + Fx2 = D : µ (2)
µ [in $/MWh] is the dual variable on the RE resource
constraint representing the marginal value of RE storage
Since (1)–(2) is a convex optimisation problem, it may bereplaced by its Karush-Kuhn-Tucker (KKT) conditions foroptimality:
0 ≤ x1 ⊥ − [A1 − (2x1 + y1)] + µ ≥ 0 (3)
0 ≤ x2 ⊥ − [A2 − (2x2 + y2)] + Fµ ≥ 0 (4)
µ free D − x1 − Fx2 = 0 (5)
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Problem Formulation and KKT Conditions
Thermal Producer
Thermal producer’s problem is to maximise its total profit,which consists of sales revenue, generation cost, and CO2
taxes:
Maximiseyj≥0 [A1 − (x1 + y1)] y1 + [A2 − (x2 + y2)] y2
−1
2C
(y21 + y2
2
)− TR (y1 + y2) (6)
Since (6) is a convex optimisation problem, it may be replacedby its KKT conditions for optimality:
0 ≤ y1 ⊥ − [A1 − (2y1 + x1)] + Cy1 + TR ≥ 0 (7)
0 ≤ y2 ⊥ − [A2 − (2y2 + x2)] + Cy2 + TR ≥ 0 (8)
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Interior Solutions
Perfect Competition
xPC1 =
F (FA1 − A2) + D + F (F − 1)RT/C
1 + F 2(9)
xPC2 =
A2 − FA1 + FD − (F − 1)RT/C
1 + F 2(10)
yPC1 =
A1 + FA2 − D − F (F − 1)RT/C −(1 + F 2
)RT
(1 + F 2) (C + 1)(11)
yPC2 =
F (A1 + FA2 − D) + (F − 1)RT/C −(1 + F 2
)RT
(1 + F 2) (C + 1)(12)
µPC =C (A1 + FA2 − D) + (F + 1)RT
(1 + F 2) (C + 1)(13)
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Interior Solutions
Cournot Oligopoly
xCO1 =
F (FA1 − A2) (C + 1) + F (F − 1)RT
(1 + F 2) (2C + 3)+
D
1 + F 2(14)
xCO2 =
(A2 − FA1) (C + 1) − (F − 1)RT
(1 + F 2) (2C + 3)+
FD
1 + F 2(15)
yCO1 =
(2C + 3)A1 + F 2 (C + 2)A1 + F (C + 1)A2 − (2C + 3)D
(1 + F 2) (C + 2) (2C + 3)
−[2C + 3 − F + F 2 (2C + 4)
]RT
(1 + F 2) (C + 2) (2C + 3)(16)
yCO2 =
F (C + 1)A1 + (C + 2)A2 + F 2 (2C + 3)A2 − F (2C + 3)D
(1 + F 2) (C + 2) (2C + 3)
−[2C + 4 − F + F 2 (2C + 3)
]RT
(1 + F 2) (C + 2) (2C + 3)(17)
µCO =(C + 1)A1 + F (C + 1)A2 − (2C + 3)D + (F + 1)RT
(1 + F 2) (C + 2)(18)
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Equilibrium Characterisation
Interior Solution Definitions
Definition
The interior solution set for perfect competition is defined as
SPC={(x, y)
PC ∈ R4+∣∣ xPC
j > 0, yPCj > 0, j = 1, 2
}(19)
Definition
The interior solution set for Cournot oligopoly with µ < 0 is defined as
SCOµ =
{(x, y)
COµ ∈ R4+
∣∣ µCO< 0, x
COj > 0, y
COj > 0, j = 1, 2
}(20)
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Equilibrium Characterisation
Perfect Competition
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Equilibrium Characterisation
Perfect Competition
Lemma
SPC is non-empty if and only if
(i) D > F
(A2 − FA1 − (F − 1)
RT
C
)if RT ≤
C(A2 − FA1)
F − 1(21a)
(ii) D >−A2 + FA1 + (F − 1)RT/C
Fif RT >
C(A2 − FA1)
F − 1(21b)
(iii) D < A1 + FA2 − F (F − 1)RT
C− (1 + F
2)RT (21c)
(iv) RT < min
A1,A2
F +F−1C
. (21d)
Furthermore, µPC > 0 is in SPC .
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Equilibrium Characterisation
Cournot Oligopoly
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Equilibrium Characterisation
Cournot Oligopoly
Lemma
SCOµ is non-empty if and only if
(i) D >(C + 1)(A1 + FA2) + RT (1 + F )
2C + 3(22a)
(ii) D <A1(C + 2)(1 + F2) + (A1 + FA2)(C + 1) − RT
[(2C + 3)(1 + F2) + F (F − 1)
]2C + 3
(22b)
(iii) RT <A1
2. (22c)
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Comparative Statics
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of Carbon Policy
Main Results
Proposition
∂xPC1
∂T> 0,
∂xPC2
∂T< 0,
∣∣∣∣∂xPC1
∂T
∣∣∣∣ > ∣∣∣∣∂xPC2
∂T
∣∣∣∣ ,∂yPC
1
∂T< 0,
∂yPC2
∂T> 0 if C <
F − 1
1 + F 2,
∂µPC
∂T> 0.
Proposition
∂xCO1
∂T> 0,
∂xCO2
∂T< 0,
∣∣∣∣∂xCO1
∂T
∣∣∣∣ > ∣∣∣∣∂xCO2
∂T
∣∣∣∣ ,∂yCO
1
∂T<
∂yCO2
∂T< 0,
∂µCO
∂T> 0.
Proposition
Off-peak RE generation under CO exceeds that under PC only if
T < (A2−FA1)(C+2)CR(F−1)(C+3)
.
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of Carbon Policy
Decomposition
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Availability
Main Results
Proposition
∂xPC2
∂D>
∂xPC1
∂D> 0,
∂yPC2
∂D<
∂yPC1
∂D< 0,
∂µPC
∂D< 0.
Proposition
∂xCO2
∂D>
∂xCO1
∂D> 0,
∂yCO2
∂D<
∂yCO1
∂D< 0,
∂µCO
∂D< 0.
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Storage Efficiency
Main Results
Assuming T = 0 and any D ∈ SPC ∩ SCOµ
Proposition
∂xPC1
∂F< 0,
∂yPC1
∂F> 0,
∂µPC
∂F> 0 if F <
(D − A1) +√
A22 + (D − A1)2
A2
∂xPC2
∂F< 0,
∂yPC2
∂F> 0.
Proposition
∂xCO1
∂F< 0,
∂xCO2
∂F< 0,
∂yCO1
∂F> 0,
∂yCO2
∂F> 0,
∂µCO
∂F> 0.
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Numerical Examples
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Data for Numerical Examples
Data for Numerical Examples
Parameter Value
A1 125
A2 250
D 250
C 0.18
F 1.80
R 1.0
T 20
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of Carbon Policy
Marginal Value of RE Storage
0 5 10 15 20 25-20
-10
0
10
20
30
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of Carbon Policy
RE Production
0 5 10 15 20 2540
60
80
100
120
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of Carbon Policy
Thermal Production
0 5 10 15 20 250
20
40
60
80
100
120
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of Carbon Policy
Producer Profits
0 5 10 15 20 250
2000
4000
6000
8000
10000
12000
14000
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of Carbon Policy
Carbon Emissions and Social Welfare
0 5 10 15 20 2560
80
100
120
140
160
180
200
0 5 10 15 20 253.3
3.4
3.5
3.6
3.7
3.8104
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Availability
Marginal Value of RE Storage
220 240 260 280 300-30
-20
-10
0
10
20
30
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Availability
RE Production
220 240 260 280 30040
60
80
100
120
140
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Availability
Thermal Production
220 240 260 280 3000
20
40
60
80
100
120
140
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Availability
Producer Profits
220 240 260 280 3000
2000
4000
6000
8000
10000
12000
14000
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Availability
Carbon Emissions and Social Welfare
220 240 260 280 30060
80
100
120
140
160
220 240 260 280 3003.3
3.4
3.5
3.6
3.7
3.8104
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Storage Efficiency
Marginal Value of RE Storage (PC)
1 1.2 1.4 1.6 1.8 2
10
15
20
25
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Storage Efficiency
Marginal Value of RE Storage (CO)
1 1.2 1.4 1.6 1.8 2-100
-80
-60
-40
-20
0
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Storage Efficiency
RE Production (T = 0)
1 1.2 1.4 1.6 1.8 20
50
100
150
200
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Storage Efficiency
RE Production (T = 20)
1 1.2 1.4 1.6 1.8 250
100
150
200
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Storage Efficiency
Thermal Production (T = 0)
1 1.2 1.4 1.6 1.8 20
20
40
60
80
100
120
140
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Storage Efficiency
Thermal Production (T = 20)
1 1.2 1.4 1.6 1.8 20
20
40
60
80
100
120
140
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Storage Efficiency
Producer Profits (T = 0)
1 1.2 1.4 1.6 1.8 20
2000
4000
6000
8000
10000
12000
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Storage Efficiency
Producer Profits (T = 20)
1 1.2 1.4 1.6 1.8 20
2000
4000
6000
8000
10000
12000
14000
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Impact of RE Storage Efficiency
Carbon Emissions and Social Welfare (T = 20)
1 1.2 1.4 1.6 1.8 220
40
60
80
100
120
140
1 1.2 1.4 1.6 1.8 23.3
3.4
3.5
3.6
3.7
3.8
3.9104
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Conclusions
Introduction Mathematical Formulation Comparative Statics Numerical Examples Conclusions
Conclusions
Summary
Transformation of power sector with higher RE penetrationentices more storage operations
Examine the impacts of carbon policy, RE availability, and
storage efficiency on market equilibria
Carbon tax may reverse results from the literature about the
role of market power in shifting RE deployment and the marginal
value of RE storage
With relatively efficient thermal production, carbon taxation
may even increase thermal output in peak period under perfect
competition
More RE availability benefits the RE producer and welfare at the
expense of the thermal producer
Higher device efficiency is more desirable for society but may be
resisted by RE producers
A carbon tax aligns private and social incentives more effectively
Future work: storage investment, transmission constraints,
intermittent RE output