IEEE POWER & ENERGY SOCIETY SECTION

Received December 5, 2019, accepted December 23, 2019, date of publication January 13, 2020, date of current version January 23, 2020.
Digital Object Identifier 10.1109/ACCESS.2020.2966224

Joint Planning and Operation for Renewable-

Storage Under Different Financial Incentives
and Market Mechanisms
KUNPENG TIAN 1, WEIQING SUN 2, DONG HAN2 , AND CE YANG1
1 Department of Control Science and Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China
2 Department of Electrical Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China

Corresponding author: Weiqing Sun (sidswq@163.com)

This work was supported by the National Natural Science Foundation of China under Grant 51777126.

ABSTRACT In this paper, a novel joint planning framework is proposed to coordinate the investment
and operation of renewable energy sources and energy storage systems (ESS) in energy and ancillary
services markets. Based on this framework, coordinated planning and operation model under centralized
and deregulated market mechanism is studied, and multiple factors such as siting and sizing of wind turbine
and ESS, the efficiency of ESS, transmission lines constraints are considered. For the centralized market
mechanism, the coordinated model aiming at maximizing social welfare is established, which is a tractable
single-level optimization problem. For the deregulated market mechanism, the coordinated model aiming at
maximizing investment profits is established, which is an intractable bi-level optimization problem as the
locational marginal price in the objective function. The bi-level optimization problem is reformulated into a
mathematical problem with equilibrium constraints by Karush–Kuhn–Tucker conditions. The big-M method
and strong dual theory are used to deal with the nonlinearity in constraints and objective functions, and
the problem is transformed into a mixed-integer linear programming, which can be solved by commercial
software. Furthermore, the impact of production tax credit and investment tax credit financial incentive
policies on investment behavior has been studied, and the evaluation indexes of electrical information and
economy have been established. The proposed approach has been implemented on the IEEE 6-bus and the
IEEE 30-bus test systems, and results justify the efficiency of the model proposed.

INDEX TERMS Investment and operation, market mechanism, mathematical problem with equilibrium
constraints, Karush–Kuhn–Tucker conditions, financial incentive.

NOMENCLATURE PARAMETERS
SETS AND INDICES bi Linear coefficient of TG cost at node i
T Set of all hours. ($/MWh).
N Set of all nodes. cug,i Regulation up cost of TG at node i ($/MWh).
L Set of all transmission lines. cdg,i Regulation down cost of TG at node i
Ng Set of thermal generator (TG) nodes. ($/MWh).
Nw Set of wind turbine (WT) nodes. max
urg,i Maximum regulation up capacity of TG at
Nes Set of energy storage systems nodes. node i (MW).
Nd Set of load nodes. max
drg,i Maximum regulation down capacity of TG
t Index of time slot. at node i (MW).
i Index of nodes. pmin
g,i Minimum power production of TG at node i
l Index of transmission line. (MW).
pmax
g,i Maximum power production of TG at node
The associate editor coordinating the review of this manuscript and i (MW).
approving it for publication was Pierluigi Siano . cbw Amortized expansion cost of WT ($/MW)

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see http://creativecommons.org/licenses/by/4.0/
13998 VOLUME 8, 2020
K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms

scw Capacity of single WT unit (MW). to push renewable power installation or reduce greenhouse
ces,d Discharging cost of ESS ($/MWh). gas emissions [2]. The European Council and the European
ces,c Charging cost of ESS ($/MWh). Union Parliament recently approved the renewable energy
cues,i Regulation up cost of TG at node i ($/MWh). directive targets that RES share in the portfolio of power
cdes,i Regulation down cost of TG at node i ($/MWh). consumption by the year 2020 reaches 20% [3]. The Republic
cbes Amortized expansion cost of ESS ($/MWh) of Ghana aims to develop and utilize RES and energy effi-
ηd The efficiency of ESS during discharging ciency technologies to achieve a 10% penetration of national
ηc The efficiency of ESS during charging electricity production by 2020 [4]. In general, the planning
sces Power capacity of single ESS unit (MW). and operation of the power system with RES recently has
ssoc
es Energy capacity of single ESS unit (MWh). been drawing extensive attention in the power industry.
ptd,i Load demands at node i during the t th period The electricity sector faces challenges from both load
(MW). demand and renewable energy target (RET) under the RPS
Bij Susceptance of the line between bus i and j framework [5]. Firstly, the load demand is considered to
(mho). continue to increase as the economy and population growth.
pmax
ij Capacity of the line between bus i and j (MW). Secondly, it is mandatory to achieve the RET in response
rd Regulation requirement as a percentage of total to the national energy strategy. How to achieve RET while
loads (%) ensuring the economy of a power system is the most con-
cerned problem of the electricity sector. We can analyze this
VARIABLES problem from three aspects: the inherent characteristics of the
ptg,i Power production of thermal generator at node i power system, the perspective of decision-makers, and the
during the t th period (MW) participation of the government.
t
urg,i Regulation up capacity provided by thermal For a complex power system, coordinated planning and
generator at node i during the t th period (MW). operation can improve the economy, because the interaction
t
drg,i Regulation down capacity provided by thermal of sources, grid, and load are considered [6]. Reference [7]
generator at node i during the t th period (MW). presents a flexible, reliable, and renewable power system
nw,i The number of WT units installed on node i. resource planning approach to coordinate generation, trans-
ptw,i Power production of the WT at node i during the mission, and energy storage systems (ESS) expansion plan-
t th period (MW). ning in the presence of demand response. Reference [8]
ptes,d,i Power discharging of ESS at node i during the presents a scenario-based stochastic active distribution net-
t th period (MW). work planning model considering the multi-type distributed
ptes,c,i Power charging of ESS at node i during the t th generation and ESS. Reference [9] presents a two-stage min-
period (MW). max-min model for jointly optimize the system’s resources
t
ures,i Regulation up capacity provided by ESS at node for meeting the RET under high security standards and renew-
i during the t th period (MW). able uncertainty. These studies have proved that it is reason-
t
dres,i Regulation down capacity provided by ESS at able to achieve the RET efficiently through the coordinated
node i during the t th period (MW). planning and operation technology of the power system.
nes,i The number of ESS units installed on node i. From the perspective of decision-makers, the impact of
etes,i State of charge (SOC) for ESS at node i during market mechanisms on decision-making programs cannot
the t th period. be ignored [10]. Under the centralized market mechanism,
θit Phase angle at node i during the t th period the independent system operators (ISO) always pursue to
λ Dual variables or Lagrange multipliers corre- maximize social welfare or the sum of all utilities while ensur-
sponding to lower bound constraints. ing the safe and reliable operation of the power system, where
λ̄ Dual variables or Lagrange multipliers corre- the social welfare is the total benefit accrued from all con-
sponding to upper bound constraints. sumption minus the total cost of generation [11]. Under the
deregulated market mechanism, market participants always
I. INTRODUCTION pursue of the maximization the expected net present value of
A. BACKGROUND AND MOTIVATION the profit obtained [12]. What is the trend of RES installed
Renewable energy sources (RES) must be popularized in capacity under the two market mechanisms of maximizing
the mid-century to limit global warming to 2 ◦ C and avoid social welfare and maximizing investment profit? Which
dangerous energy shortages and climate change [1]. It is market mechanism can push the development of RES? These
encouraging that a series of policies on promoting RES devel- problems will be solved in this paper.
opment have been implemented around the world to accel- The visible hand of government is as important as the
erate the transition to clean energy generation. Renewable invisible hand of market competition [13]. Financial incen-
portfolio standard (RPS), which makes sure a proportion of tives to reduce CO2 emissions are studied in [14] and they
RES included in the final energy consumption mix, is a com- can promote the development of RES and achieve RET.
mon policy instrument employed in many countries as a tool RES power generation is characterized by high capital costs

VOLUME 8, 2020 13999

 K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms
together with weather-dependent production [15]. As a result,
these RES are prevented from being perfectly competitive in
a liberalized electricity market. The incentive of RES invest-
ment through financial subsidy policies is widely recognized
in many countries. Reference [16] divides financial subsidy
policy into short-term (i.e. direct subsidies, an income tax
exemption for RES) and long-term (i.e. feed-in tariffs, auc-
tions, and the quota system). In [17], financial incentives
are classified into quantity-based and price-based, and where
support schemes for the former are determined by the market,
support schemes for the latter are characterized by a fixed
price determined by the regulator. In general, RES investment
has characteristics of a substantial initial investment and high
risk, a long return period, and investors have difficulties
receiving financing [18]. Therefore, RES investment prob-
lems with the RPS framework typically call for government
interventions to address market failures associated with target
and investment in RES [19].
Existing substantial literatures use different techniques to
explore the planning and operation of power systems from
different aspects [20]. The impact of market mechanism
and financial incentives on TG expansion planning are stud-
ied in [21]. Different from the existing research, this paper
focuses on the coordinated planning of RES and ESS under
the framework of RPS. The coordinated planning model
with the objective function of maximizing social welfare and
maximizing investment profits is respectively established to
study RES investment under the centralized and deregulated
market mechanism. The potentialities and fragilities of finan-
cial incentives for RES development are analyzed. The coor-
dinated planning method that can determine the siting and
sizing of RES and ESS in this paper is a static model, which
only determines where and what type of RES and ESS should
be built in the extended power system. It should be noted that
time series production is used to describe the dynamic bal-
ance characteristics of the power system to achieve the coor-
dination of planning and operation scheduling [22]. The static
model and time series production are combined to constitute
the tradeoff between accuracy and tractability. As mentioned
in [23], the proposed coordinated planning approach can be
simply extended to become a dynamic tool at the expense of
a higher computation burden.
B. CONTRIBUTIONS
The main contributions of the paper can be summarized as
follows:
1) The coordinated planning and operation model of RES
and ESS with co-optimization of energy and auxiliary ser-
vices is presented respectively under the centralized and
deregulated market mechanism. The impact of financial
incentives is studied, the potentialities and fragilities of these
subsidy policies are analyzed. Our research results provide
suggestions for the existing financial incentives to promote
the development of RES and ESS.
2) For the centralized market mechanism, the total revenue
collected from the participation of RES and ESS in electricity
14000
markets and financial policies does not exceed the financial
subsidy. However, the total revenue under deregulated market
mechanisms is no less than the financial subsidy.
C. PAPER ORGANIZATION
The remainder of this paper is organized as follows. Section II
describes the mathematical model of generator and ESS.
In Section III, a detailed coordinated planning and operation
model is provided under the centralized market mechanism.
In Section IV, a bi-level coordinated planning and operation
model is presented under the deregulated market mechanism.
In Section V, the influence of financial incentive policy on
investment behavior is analyzed. In Section VI, the pro-
posed model is tested on the modified IEEE 6-bus and IEEE
30-bus test system and the obtained numerical results are
discussed extensively. Finally, relevant conclusions are drawn
in Section VII.
II. MATHEMATICAL MODELING
Without losing the general, the power system to be expanded
contains N buses and L transmission lines. Since the focus in
this paper is on coordinated planning and operation of RES
and ESS, investors in RES and ESS are assumed to belong to
one coalition. As a daily time horizon can sufficiently model
the dynamics and cyclic behaviors of ESS, and a daily time
is divided into T = 24 hourly time slots. Considering the
computational burden, we assume that future demand and
supply information is known.
A. THERMAL GENERATOR
The traditional power supply represented by TG. The oper-
ation cost of TG includes the generation cost or fuel con-
sumption cost, which can be expressed as a quadratic model.
Besides, the regulation cost of auxiliary services provided
by TG is considered. The total costs and constraints are as
follows [24].
XX 
bi ptg,i + cug,i urg,i
t
+ cdg,i drg,i
t
(1)
t∈T i∈Ng
t
0 ≤ urg,i max
≤ urg,i : λtg,ur,i , λ̄tg,ur,i (2)
t
0 ≤ drg,i max
≤ drg,i : λtg,dr,i , λ̄tg,dr,i (3)
pmin t t max
g,i + drg,i ≤ pg,i ≤ pg,i − urg,i
t
: λtg,i , λ̄tg,i (4)
Equation (1) represents the operation and regulation cost of
all TG on the daily time horizon. Equations (1) and (3) denote
the ramp-up and ramp-down capacity of TG, respectively.
Equation (4) indicates that the power production of TG is
restricted by the minimum and maximum limits. For nota-
tional simplicity, we assume that each node has a TG. In fact,
the power production of the node without TG is always zero,
and the WT and ESS are similar, which will not be discussed
later.
B. RENEWABLE ENERGY SOURCES
RES represented by WT and photovoltaic has developed
rapidly in recent years. Although their production power
VOLUME 8, 2020
K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms

characteristics vary greatly, there is little difference in the
mathematical model, and this paper takes wind power as
an example. The investment cost of WT (per MW capacity)
prorated capital cost amortized over its life span. It is widely
known that RES generation has zero marginal cost [25], and
the production cost of WT is ignored. The total cost and
operational constraints of WT are as follows.
X by lower bound and upper bound to avoid deep discharging
cbw nw,i (5)
i∈Nw
0 ≤ ptw,i ≤ scw nw,i : λtw,i , λ̄tw,i
nw,i ≥ 0 : λn,w,i
exactly relaxed to a linear model to reduce the computational
burden [29]. Equation (12) bounds that the discharging power
and regulation up the capacity of ESS must be within the
rated power. Equation (13) limits that the charging power and
regulation down the capacity of ESS must be within the rated
power. Equation (14) represents the dynamics of the stored
energy. Equation (15) indicates that the SOC of ESS is limited
and overcharging. Equation (16) binds the initial and final
stored energy levels of each ESS in every representative day.
(6) Equation (17) forces the number of ESS to be non-negative,
(7) and nes,i can also be considered as a continuous variable.

Equation (5) represents the investment cost of WT.
Equation (6) denotes the power production limit of WT.
Equation (7) indicates the number limit of WT is a positive
number. Note that nw,i is considered as a continuous variable
instead of an integer variable to reduce the computational
burden due to the capacity of a single WT unit is very small.
C. ENERGY STORAGE SYSTEM
For a power system integrated with RES generation, the ESS
plays a critical role in mitigating the system uncertainties and
maintaining the supply and demand balance. The total cost of
ESS includes operation and maintenance (O&M) cost, regu-
lation cost, and construction investment [26]. The frequent
charging and discharging cycling of ESS lead to extra O&M
costs as it may accelerate ESS degradation, and the O&M
cost is usually modeled to be proportional to the amount of
charging and discharging energy of ESS [27].
XX
ces,d ptes,d,i /ηd + ces,c ηc ptes,c,i
t∈T i∈Nes
 X
+ cues,i ures,i
t
+ cdes,i dres,i
t
+ cbes nes,i (8)
i∈Nes
0 ≤ ptes,c,i ≤ sces nes,i : λtes,c,i , λ̄tes,c,i (9)
0 ≤ ptes,d,i ≤ sces nes,i : λtes,d,i , λ̄tes,d,i
(10)
III. COORDINATED PLANNING AND OPERATION UNDER
CENTRALIZED MARKET MECHANISM
In this section, a coordinated planning and operation model
is presented under the centralized market mechanism. The
centralized market mechanism endows ISO with full physical
rights, and the operation strategy of WT and ESS is obtained
by solving the multi-stage economic scheduling problem.
Therefore, the coordinated planning model aiming at max-
imizing social welfare is proposed to determine the siting
and sizing of WT and ESS. Maximizing social welfare is
one of the most important economic features of market-based
power systems. A maximization social welfare problem can
be solved in a minimization form as the total cost [30].
A. MAXIMIZING SOCIAL WELFARE
The social optimum scheme is the solution to the follow-
ing optimization problem with an objective to minimize
the overall system cost including the total operation cost,
the regulation cost, the investment cost. The dynamic balance
constraints of power system with multi-time coupling are
considered, such as the operation constraints of TG and WT,
dynamic constraints of ESS, node power balance, line power
flow constraints, and regulation capacity constraints.
XX 
min bi ptg,i +cug,i urg,i
t
+cdg,i drg,i
t

ptes,c,i ptes,d,i = 0 (11) t∈T i∈Ng

0 ≤ ptes,d,i /ηd + ures,i t
≤ sces nes,i : λtes,ur,i , λ̄tes,ur,i
XX
(12) + ces,d ptes,d,i /ηd + ces,c ηc ptes,c,i
0 ≤ ηc ptes,c,i + dres,i
t
≤ sces nes,i : λtes,dr,i , λ̄tes,dr,i (13) t∈T i∈Nes
 X X
es,i + ηc pes,c,i 1t − pes,d,i /ηd 1t
etes,i = et−1 t t + cues,i ures,i
t
+cdes,i dres,i
t
cbw nw,i + cbes nes,i


(14) +
es nes,i − dres,i : λes,r,i , λ̄es,r,i
t
ures,i ≤ etes,i ≤ ssoc t t t
(15) i∈Nw i∈Nes
X (18)
ptes,d,i /ηd − ηc ptes,c,i = 0 : λtes,s,i


(16)
t∈T The first sum calculates the cost of operation and ancil-
nes,i ≥ 0 : λn,es,i (17) lary services of TG. The second sum calculates the cost of
discharging, charging, and regulation of ESS. The third and
Equation (8) stands for the sum of degradation cost, regula- fourth sum represent the investment cost of WT and ESS,
tion cost and investment cost of ESS. Equations (9) and (10) respectively.
denote the maximum charging and discharging limits of ESS.
ptg,i + ptw,i + ptes,d,i − ptes,c,i


Equation (11) ensures that no ESS simultaneously charges  
and discharges which results in mixed-integer linear pro-
X
− ptd,i = Bij θit − θjt : λtn,i (19)
gramming (MILP) model for ESS in some references [28]. (i,j)∈L
Actually, Equation (11) is redundant when charging and dis-  
charging efficiency is considered and the MILP model can be −pmax
ij ≤ Bij θit − θjt ≤ pmax
ij : λtij , λ̄tij (20)

VOLUME 8, 2020 14001

 K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms

X X X
t
urg,i + t
ures,i ≥ rd ptd,i : λtur (21) the power system regulation up and down prices. The third
i∈Ng i∈Nes i∈Nd sum represents the operation cost of discharging, charging,
X X X as well as providing ancillary services of ESS. The fourth
t
drg,i + t
dres,i ≥ rd ptd,i : λtdr (22)
and fifth sum represent the investment cost of WT and ESS,
i∈Ng i∈Nes i∈Nd
respectively. LMP, system regulation up and down prices
Equation (19) represents the nodal power balance. are Lagrange multiplier (dual variables) associated with the
Equation (20) restricts the power flow of the specified trans- constraints (19), and (21)-(22), respectively. A theorem can
mission line considering its maximum thermal limit, and be obtained:
based on the standard DC power flow assumptions of lossless Theorem: The coordinated planning of WT and ESS
lines, constant bus voltages, and small phase angle difference investment with maximizing social welfare always yields
between neighboring buses in line with the previous research zero investment profit if the initial state of charge at each
works in the area, such as [31]. Equation (21) represent ESS is zero. No matter WT or ESS, the revenue collected
the minimum requirement on the regulation up capacity. from participating in energy and ancillary services markets
Equation (22) represent the minimum requirement on the exactly covers the cost of investment and operation at a social
regulation down capacity. Besides, TG operation constraints optimum. The balance between WT income and investment
are given in (1)-(4). The investment and operation constraints is proved by applying the generalized Lagrange multiplier
of WT are given in (6)-(7). The investment and operation method, first-order optimality conditions and complementary
constraints of ESS are given in (9)-(17). slackness conditions [34].
The coordinated planning for maximizing social welfare Proof: The Karush–Kuhn–Tucker (KKT) conditions are
is formulated as MILP, which can be solved by commer- obtained by differentiating the Lagrangian function by the
cial software (i.e. Gurobi solver). The investment decision primal variables. The detailed KKT conditions are given in
variables are the siting and sizing of WT and ESS. The Appendix A, part of which is as follows.
operation decision variables the power generation of TG and −λtn,i − λtw,i + λ̄tw,i = 0 (24)
WT, the power discharging and charging as well as SOC of X
ESS, the voltage phase angles, and the regulation up/down cbw − λn,w,i − scw λ̄tw,i = 0 (25)
capacity provided by TG or ESS. t∈T
Equations (24) and (25) are obtained by differentiating the
B. ZERO INVESTMENT PROFITS Lagrangian function by the ptw,i and nw,i , respectively. The
In the investment case of maximizing social welfare under the revenue of WT production can be derived through arithmetic
centralized market mechanism, ESS straightforwardly profits manipulation of the KKT conditions. We multiply both sides
through intertemporal arbitrage, also known as load shifting of equation (24) by ptw,i , and obtain equation (26).
and peak shaving [32]. The total revenue collected from XX XX
λtn,i ptw,i = λ̄tw,i ptw,i − λtw,i ptw,i


(26)
ancillary services provided by ESS and power production
t∈T i∈Nw t∈T i∈Nw
by WT. Note that in this study, the storage facility is part
of a storage-wind nor storage-solar a coalition, instead of is To simplify equation (26), we apply the complementary
part of the portfolio of a dominant generation company. The slackness conditions equation (27) and (28) of equation (6).
profits of coalition are defined as the difference between the λtw,i ptw,i = 0 (27)
total revenue and the investment and operation costs. Let TIP
λ̄tw,i ptw,i − scw nw,i = 0


(28)
represent the total investment profits of the coalition.
XX XX Substituting equation (27) and (28) into (26), and obtain the
max λtn,i ptw,i + λtn,i ptes,d,i − λtn,i ptes,c,i revenue of WT generation from energy market.
t∈T i∈Nw t∈T i∈Nes XX
XX t c
XX λ̄tw,i ptw,i − λtw,i ptw,i = λ̄w,i sw nw,i (29)
+ λtur ures,i
t
λtdr dres,i
t
ces,d ptes,d,i /ηd


+ − t∈T i∈Nw t∈T i∈Nw
t∈T i∈Nes
 Similar, we multiply both sides of equation (25) by nw,i , and
+ ces,c ηc ptes,c,i + cues,i ures,i
t
+ cdes,i dres,i
t
obtain equation (30).
X X X XX
cbw nw,i − cbes nes,i cbw nw,i = λ̄tw,i scw nw,i + λn,w,i nw,i


− (23) (30)
i∈Nw i∈Nes i∈Nw t∈T i∈Nw

The first sum calculates the payment the WT receives To simplify equation (30), we apply the complementary
from the energy market that settles in the LMP widely used slackness conditions of equation (7).
in the wholesale electricity market in USA [26]. The first λn,w,i nw,i = 0 (31)
two term in the second sum calculates the payment the ESS
receives from the energy market that settles in the LMP. Substituting equation (30) into (29), and obtain equation (32).
XX
XX t c
The third and fourth term in the second sum calculates the λ̄tw,i scw nw,i +λn,w,i nw,i = λ̄w,i sw nw,i (32)
payment of the ESS from the regulation market that settles in t∈T i∈Nw t∈T i∈Nw

14002 VOLUME 8, 2020

K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms

Compare equation (32) with equation (33), we conclude that and the upper and lower level problems are interactive as
the revenue of WT generation is always zero. Similarly, shown in Figure 1.
the revenue and investment of ESS are balanced, which is
proved in reference [33].

IV. COORDINATED PLANNING AND OPERATION UNDER

DEREGULATED MARKET MECHANISM
Different from the centralized market mechanism in
Section III, the coordinated planning and operation model
in this section is proposed under the deregulated market
mechanism. The deregulated market mechanism endows
investors with full physical rights and allows the owner to
directly control ESS and WT operation for every node and
at every period [10]. Therefore, the coordinated planning and
operation model aiming at maximizing investment profit is
proposed to determine the siting and sizing of WT and ESS.

FIGURE 1. An illustration of the bi-level program and the interfaces

A. MAXIMIZING INVESTMENT PROFIT
between the upper-level and lower-level problems.
Following the convention in the related reference [35],
we consider the profit-maximization behavior of a wind-
storage coalition. The optimal siting and sizing of WT and Let X and Y represent the decision variables of upper-level
ESS for profit-maximizing investment are formulated as and lower-level problems respectively. As shown in Figure 1,
bi-level problems. The upper-level problem models the plan- for a given decision variable X0 of the upper-level problem,
ning and operation decisions of the WT and ESS facility the decision variable Y and the Lagrange multiplier or dual
from the coalition’s perspective. The lower-level problems variable of constraints are obtained by solving the lower-level
represent the market clearing process, for each hour of a day. market clearing problem and returned to the upper level. It is
The upper-level problem of maximizing TIP is as follows: not difficult to find that an optimization problem (upper-level
XX XX problem) contains another optimization problem (lower-level
max λtn,i ptw,i + λtn,i ptes,d,i − λtn,i ptes,c,i problem) as a constraint [36]. The problem of maximizing
t∈T i∈Nw t∈T i∈Nes investment profit under the deregulated market mechanism
XX is a classic bi-level optimization. In the profit-maximizing
+ λtur ures,i
t
λtdr dres,i
t
ces,d ptes,d,i /ηd


+ −
investment case, investment and operation of WT and ESS
t∈T i∈Nes
 are free, and the wind-storage coalition has positive benefits.
+ ces,c ηc ptes,c,i + cues,i ures,i
t
+ cdes,i dres,i
t
However, the bi-level optimization problem is intractable
X X and cannot be solved directly by commercial software. The
− cbw nw,i − cbes nes,i
solution method is given in Section IV-B.
i∈Nw i∈Nes
s.t. (6) − (7) , (9) − (17) (33)
B. SOLUTION METHOD
Equation (33) represents a tractable MILP or even LP (if the To solve the bi-level optimization in Section IV-A, the inher-
WT and ESS capacity is relaxed to continuous variables). The ent nonlinearity in the definition of LMP and non-convex
decision variables of the upper-level problem are the invest- nonlinearity in the objective function described by equa-
ment and operation strategy of WT and ESS. The lower-level tion (33) need to be tackled [37].
problem of market clearing are as follows: Firstly, the bi-level optimization is recast as a single-
XXn o level linearized mathematical problem with equilibrium con-
min bi ptg,i + cug,i urg,i
t
+ cdg,i drg,i
t
straints (MPEC). This is achieved by substituting the convex
t∈T i∈Ng lower-level problem with an equivalent set of constraints to
XXn the upper-level problem [38]. This set of constraints consists
+ ces,d ptes,d,i /ηd + ces,c ηc ptes,c,i
of KKT conditions of a lower-level problems, which can
t∈T ∈Nes
o rewrite the convex optimization of lower-level problem into
+ cues,i ures,i
t
+ cdes,i dres,i
t
the constraint condition of the upper-level problem. The KKT
s.t. (2) − (4) , (19) − (22) (34) conditions for the lower-level problem (34) are as follows.

The decision variables of the lower-level problem are the bi − λtn,i − λtg,i + λ̄tg,i = 0 (35)
scheduling of TG and the phase angle of nodes. The nodal
prices and the regulation up/down prices in the upper-level cug,i − λtg,ur,i + λ̄tg,ur,i + λ̄tg,i − λtur =0 (36)
problem are the optimal solution of the lower-level problem, cdg,i − λtg,dr,i + λ̄tg,dr,i + λtg,i − λtdr =0 (37)

VOLUME 8, 2020 14003

 K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms

    
 λ̄t ≤ 1 − k̄ t M
X
Bij λtn,i − λtn,j ij ij
j∈N
  (57)
 pmax − Bij θ t − θ t ≤ k̄ t M
X   ij i j ij
+ Bij λ̄tij − λtij + λtji − λ̄tji = 0 (38) 
 λur ≤X1 − kur X
t t


j∈N
 M
X
t t t t (58)
λg,ur,i urg,i =
t t
0 (39)  −r d pd,i + ur g,i + ures,i ≤ kur M
   i∈Nd i∈Ng i∈Nes
λ̄tg,ur,i urg,i
t max
− urg,i =0 (40) 
 λdr ≤X1 − kdr X
t t


 M
λtg,dr,i drg,i
t
=0 (41) t t
X
t t (59)
   −rd pd,i + drg,i + dres,i ≤ kdr M
λ̄g,dr,i drg,i − drg,i
t t max 
=0 (42) i∈N d i∈N i∈N g es
  k tg,ur,i , k̄g,ur,i
t
, k tg,dr,i , k̄g,dr,i
t
,
λtg,i pmin t t
g,i + drg,i − pg,i = 0 (43)
  k tg,i , k̄g,i
t
, k tij , k̄ijt , kur
t
, kdr
t
∈ {0, 1} (60)
λ̄tg,i ptg,i − pmax
g,i + ur t
g,i = 0 (44)
n  o where M represents a sufficiently large constant, and which
λtij −pmaxij − B ij θ t
i − θ t
j =0 (45) are used to form disjunctive constraints [40]. So far, the inher-
n   o ent nonlinearity of in the definition of LMP is solved, and
λ̄tij Bij θit − θjt − pmax ij =0 (46) optimization problem (34) has been reformulated as a set of
X X X linear equality (35)-(38) and inequality constraints (50)-(60)
λtur rd ptd,i − λtur t
urg,i − λtur t
ures,i =0 (47)
over continuous and binary variables, and the bi-level opti-
i∈Nd i∈Ng i∈Nes
X X X mization problem is rewritten as single-level optimization.
λtdr rd ptd,i − λtdr t
drg,i − λtdr t
dres,i =0 (48) Secondly, the nonlinearity of the objective function in
i∈Nd i∈Ng i∈Nes equation (33) can be tackled through arithmetic manipulation
λtg,ur,i , λ̄tg,ur,i , λtg,dr,i , λ̄tg,dr,i , λtg,i , and the strong duality theory. For a given upper-level decision
λ̄tg,i , λtn,i , λtij , λ̄tij , λtur , λtdr ≥ 0 (49) variable X0 , we define the following primal lower-level (PL)
optimization problem as Q(X0 , Y ).
In addition, the KKT condition also includes the primal
constrain (1)-(4), and (19)-(22). Solving optimization prob- PL : min C 0 X0 + D0 Y
lem (34) is equivalent to solving the nonlinearity equations s.t. JY ≥ K
(35)-(49) as optimization problem (34) is convex and Slater’s
EX0 + FY ≥ G
condition holds. Note that the nonlinearity is due to the
complementary slackness constraints in equations (39)-(48). Y ≥0 (61)
Nonlinear complementary slackness constraints can be where C and D are the coefficients of upper-level and lower-
reformulated into linear constraints by big-M method. Alter- level decision variables respectively. J and K are the left-hand
natively, one could employ the approach of Schur’s decom- and right-hand coefficients of equations (1)-(4). E and F are
position, which is based on SOS-1 variables [39]. the left-hand coefficients of equations (19)-(22), and G is the
right-hand coefficients. We can write dual lower-level (DL)
(  
λtg,ur,i ≤ 1 − k tg,ur,i M
(50) optimization problems as follows.
t
urg,i ≤ k tg,ur,i M
(   DL : max C 0 X0 + K 0 5 + (G − EX0 )0 9
λ̄tg,ur,i ≤ 1 − k̄g,ur,i t
M s.t. J 0 5 + F 0 9 ≤ D
(51)
max t t
urg,i − urg,i ≤ k̄g,ur,i M 5, 9≥0 (62)
(  
λtg,dr,i ≤ 1 − k tg,dr,i M where 5 and 9 are dual variables or Lagrange multipliers
(52)
drg,it
≤ k tg,dr,i M of corresponding constraints respectively. According to the
(   properties of strong duality theory, we can get equation (63).
λ̄tg,dr,i ≤ 1 − k̄g,dr,i t
M
(53) C 0 X0 + D0 Y = C 0 X0 + K 0 5 + (G − EX0 )0 9 (63)
max t t
drg,i − drg,i ≤ k̄g,dr,i M
(   The constant terms related to the upper-level decision vari-
λtg,i ≤ 1 − k tg,i M ables are ignored, and we get equation (64).
(54)
ptg,i − pmin t t XX 
g,i − drg,i ≤ k g,i M bi ptg,i + cug,i urg,i
t
+ cdg,i drg,i
t
(  
λ̄tg,i ≤ 1 − k̄g,i t
M t∈T i∈Ng
(55) X X 
max t t
pg,i + urg,i − pg,i ≤ k̄g,i t
M = −urg,i λ̄g,ur,i −drg,i
max t
λ̄g,dr,i +pmin
max t
λt
g,i g,i −p λ̄
max t
g,i g,i
t∈T i∈Ng
  
 λt ≤ 1 − k t M
ij
ij
XX XX
 (56) + ptd,i λtn,i − ptw,i λtn,i
 Bij θ t − θ t + pmax ≤ k t M
i j ij ij t∈T i∈Nd t∈T i∈Nw

14004 VOLUME 8, 2020

K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms

XX X X  
ptes,d,i −ptes,c,i λtn,i − pmax λtij + λ̄tij


− ij for auxiliary services provided by ESS. TIP represents the
t∈T i∈Nes t∈T (i,j)∈L coalition profit without subsidies. Under the ITC policy,
+ rd
XX
ptd,i λtur +λtdr

the total coalition profit is given as follow.
X X
XX
t∈T i∈Nd TCPITC = σw cbw nw,i +σes cbes nes,i + TIP (67)
t
λtur +dres,i
t
λtdr


− ures,i (64) i∈Nw i∈Nes
t∈T i∈Nes where σw and σes is a subsidies portion of the capital cost
where the left-hand side is the primal optimal objective value provided by ITC policies.
and the right-hand side is the dual optimal objective value. The coordinated planning and operation model of maxi-
We notice that the WT and ESS profits could be obtained after mizing social welfare and maximizing investment profit is
reordering. studied under the above two financial incentives. Electrical
XX XX information and economic indicators are used to evaluate
ptw,i λtn,i + ptes,d,i − ptes,c,i λtn,i


the impact of PTC and ITC financial incentive policies on
t∈T i∈Nw t∈T i∈Nes
XX investors’ behavior under two market mechanisms. Electri-
t
λtur t
λtdr


+ ures,i + dres,i cal information indicators include installed capacity of WT
t∈T i∈N
es and ESS, the amount of wind energy utilization and cur-
X X tailment. Economic indicators include social cost, profits of

= −urg,i λ̄g,ur,i −drg,i
max t
λ̄g,dr,i +pmin
max t
g,i λg,i −pg,i λ̄g,i
t max t
wind-storage alliance and financial subsidy.
t∈T i∈Ng
XX X X  
+ ptd,i λtn,i − pmax
ij λ t
ij + λ̄ t
ij VI. NUMERICAL RESULTS
t∈T i∈Nd t∈T (i,j)∈L In this section, the performance of the proposed method is
XX evaluated based on the modified IEEE 6-bus test system and

− bi ptg,i + cug,i urg,i
t
+ cdg,i drg,i
t
the IEEE 30-bus test system [43] to demonstrate the mechan-
t∈T i∈Ng
XX ics of the proposed method. The numerical examples are
ptd,i λtur + λtdr


+ rd (65) implemented in MATLAB 2018a with Gurobi as the MILP
t∈T i∈Nd solver, and run on a 3.5GHz Intel Xeon(R) Gold 6135M
The above expression is a linear function of dual vari- processor with 48 GB RAM and 64-bit Windows 10 system.
ables, and the nonlinearity in the objective function in (33) is The convergence tolerance of the algorithm is set on 10−3 .
now tackled. Finally, the problem of maximizing investment The daily 24-hour load demand and WT production profiles
returns is transformed into a MILP, which can be solved are given in Figure 2.
efficiently using commercial packages.

V. FINANCIAL INCENTIVES FOR INVESTMENT AND

OPERATION
While the cost of solar and wind energy has declined rapidly
in the recent past, it remains higher than those of conventional
energy technologies. A large number of policy instruments
have been implemented to support RES in worldwide. In [41],
the key policy instruments include feed-in-tariffs, investment
tax credits, subsidies, favorable financing, mandatory access
and purchase, renewable energy portfolio standards and
public investment was highlighted. The policy instruments
can be summarized into two categories [42]: a production
based financial incentive with a production tax credit (PTC)
for every MWh energy and every MW regulation capacity.
Besides, capacity based financial incentives with an invest-
ment tax credit (ITC) that equals a certain portion of the FIGURE 2. Load demand and WT production profiles.
capital cost. Assuming the PTC policy offers the wind-storage
coalition a tax credit, and then the total coalition profit is In the numerical example, the daily capital cost of WT
given as follows. and ESS are obtained by applying the capital recovery factor.
The amortized daily cost of WT and ESS are 100 $/MW and
XX XX
TCPPTC = ωw ptw,i + ωes ptes,d,i
t∈T i∈Nw t∈T i∈Nes
300 $/MWh, respectively. The efficiency of ESS is set
to 95%, and the ESS can be fully charged and discharged
+ ωes,r t t


ures,i + dres,i + TIP (66)
in 2 hours. The operation parameters are set as follows. The
where ωw and ωes are energy delivery subsidies provided by regulation requirement is set to be 10% of the system load
PTC policies for WT and ESS respectively. ωes,r is subsidy demand. The maximum regulating up and down capacity of

VOLUME 8, 2020 14005

 K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms
TG is set to 10% of the maximum capacity, and regulation
up/down the cost of TG is 5 $/MW. The degradation costs
associated with charging/discharging of ESS are 0.5 $/MWh,
and regulation up/down cost of ESS are 0.5 $/MW [33].
For financial incentive policies, we gradually increase the
intensity of PTC and ITC financial subsidies to study the
investment behavior of market participants. We set the step
length of ωw , ωes and ωes,r to 6 $/MWh, 2 $/MWh and
1 $/MW, respectively. The step length of σw and σes are set
to 8%. Finally, the energy delivery subsidy for WT gradually
increasing from 0 $/MWh to 30 $/MWh, and the energy
delivery subsidy for ESS gradually increasing from 0 $/MWh
to 10 $/MWh, and the auxiliary service subsidy for ESS
from 1 $/MW to 5 $/MW under the PTC policy. The capital
cost subsidy portion of WT and ESS gradually increasing
from 0% to 40% under the ITC policy.
A. MODIFIED IEEE 6-BUS TEST SYSTEM
This section presents the results obtained on a modified IEEE
6-bus test system. The modified test system has 6 nodes that
are connected through 7 transmission lines, all nodes are
allowed to install WT and ESS. The maximum load demand
of node 3 is 56 MW, and that of node 4 and node 5 are
112 MW. The parameter of transmission lines and TG are
given in Table 1 and Table 2.
TABLE 1. Transmission lines data of modified IEEE 6-bus test system.
TABLE 2. Thermal generator data of modified IEEE 6-bus test system.
Firstly, we study the impact of PTC financial incentives on
investors’ behavior under the centralized market mechanism.
We numerically compare the installed capacity of WT and
ESS achieved under PTC financial incentives, as well as
the amount of wind energy utilization and curtailment. The
results are given in Figure 3.
Let (ωw , ωes , ωes,r ) represent the energy delivery subsidy
for WT and ESS, and auxiliary service subsidy for ESS under
different PTC intensity. The installed capacity of WT and
ESS are non-decreasing with the increase of PTC financial
incentive intensity, as well as the amount of wind energy
14006
FIGURE 3. The impact of PTC policy intensity on total capacity and wind
energy utilization under the centralized market mechanism.
utilization and curtailment. This result verifies the driving
force of PTC policy on WT investment, and we can choose
the appropriate intensity of PTC policy to achieve the pre-
determined RET. Further, the social cost, coalition profit and
financial subsidy are given in Figure 4.
FIGURE 4. The impact of PTC policy intensity on coalition profits and
social costs under the centralized market mechanism.
The social cost is decreasing while the coalition profit is
increasing with the increase of PTC policy intensity. Wind
energy is zero marginal cost, and ESS can promote the uti-
lization of wind energy and reduce the power production of
traditional generators in peak hours. As a result, the social
cost is non-increasing. Note that the coalition profit is zero
without financial subsidy. Besides, the fact that the financial
subsidy is greater than the total coalition profit TCPPTC indi-
cates that the revenue (TIP) of WT and ESS from the energy
and auxiliary service market is non-positive.
Secondly, the impact of ITC policy on investors’ behavior
under the centralized market mechanism is studied. Similarly,
the installed capacity of WT and ESS under ITC financial
incentives is compared, as well as the amount of wind energy
utilization and curtailment. The results are shown in Figure 5.
Let (σw , σes ) represent the capital cost subsidies portion
for WT and ESS under different ITC intensity. Similar to
PTC financial incentives, ITC can also promote the installed
capacity of WT and ESS as well as the amount of wind energy
utilization. When the proportion of ITC subsidy for WT and
VOLUME 8, 2020
K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms
FIGURE 5. The impact of ITC policy intensity on total capacity and wind
energy utilization under the centralized market mechanism.
ESS reaches 40%, the utilization of wind energy increases
from 411 MWh to 956 MWh. The social cost, coalition profit
and financial subsidy are given in Figure 6.
FIGURE 6. The impact of ITC policy intensity on coalition profits and
social costs under the centralized market mechanism.
The social cost is decreasing while the coalition profit is
increasing with the increase of ITC policy intensity. However,
it should be noted that coalition profit TCPITC equals to finan-
cial subsidy under ITC policy. This means that the revenue
TIP of WT and ESS in the energy and auxiliary service market
is zero.
For the numerical results from the centralized market
mechanism, from Figure 3 and Figure 5, both PTC and ITC
financial incentives can increase the installed capacity of WT
and ESS to improve the utilization of wind energy. From
Figure 4 and Figure 6, both PTC and ITC financial incentives
can reduce the social cost and increase the coalition profit.
Besides, the revenue TIP of WT and ESS from the energy
and auxiliary service market under the centralized market
mechanism is non-positive.
Thirdly, we study the impact of PTC financial incentives on
investors’ behavior under the deregulated market mechanism.
The installed capacity of WT and ESS as well as the
amount of wind energy utilization and curtailment result
from under the deregulated market mechanism are given in
Figure 7. We find that the same intensity of PTC policy has
no significant incentive effect on WT and ESS investment.
VOLUME 8, 2020
FIGURE 7. The impact of PTC policy intensity on total capacity and wind
energy utilization under the deregulated market mechanism.
Although the PTC subsidy intensity has been from (0, 0, 0) to
(30, 10, 5), the installed capacity of WT has only increased by
11 MW (form 33 MW to 44 MW), and the installed capacity
of ESS has hardly increased, and the amount of wind energy
utilization has only increased by 105 MWh (from 404 MWh
to 509 MWh). The capital cost of WT and ESS is much higher
than the production cost of TG, which makes it difficult for
wind-storage coalition to profit from investment, and then the
incentive effect of low-intensity PTC policy is limited. The
social cost, coalition profit and financial subsidy are given
in Figure 8.
FIGURE 8. The impact of PTC policy intensity on coalition profits and
social costs under the deregulated market mechanism.
It can be seen from Figure 8 that the social cost is sta-
ble and the coalition profit is gradually increasing with the
increase of financial incentive intensity of PTC. Comparing
Figure 4 with Figure 8, the following two points are different.
Without PTC financial subsidy (0, 0, 0), the coalition profit
under centralized market mechanism is zero (Figure 4), while
the coalition profit under deregulated market mechanism is
1768 $ (Figure 8). Besides, the coalition profit TCPPTC under
deregulated market mechanism (Figure 8) is greater than
financial subsidy indicates that the revenue TIP of WT and
ESS from the energy and auxiliary service market is positive.
Finally, the impact of ITC policy on investors’ behavior
under the deregulated market mechanism is studied.
Figure 9 shows the impact of ITC financial incentive
14007
 K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms
FIGURE 9. The impact of ITC policy intensity on total capacity and wind
energy utilization under the deregulated market mechanism.
intensity on installed capacity of WT and ESS under deregu-
lated market mechanism, as well as the impact on amount
of wind energy utilization and curtailment. The financial
incentive intensity of ITC increases from (0, 0) to (40%,
40%), the installed capacity of WT and ESS increased by
56 MW (form 33 MW to 89 MW) and 16 MW (form
6 MW to 22 MW) respectively, and the amount of wind
energy utilization increased by 509 MWh (from 404 MWh
to 913 MWh). The social cost, coalition profit and financial
subsidy are given in Figure 10.
FIGURE 10. The impact of ITC policy intensity on coalition profits and
social costs under the deregulated market mechanism.
ITC financial incentive policies that directly reduce capital
costs can effectively improve the installed capacity of WT and
ESS and reduce social costs. When ITC financial incentive
subsidy reaches (40%, 40%), social cost is reduced by 5574 $
(from 91016 $ to 85442 $), and total coalition profit increases
by 12919 $ (from 1768 $ to 14687 $). Besides, the impact of
reserve capacity on coalition profits and social costs under the
different market mechanisms is analyzed, and the results are
given in Figure 11.
Reserve capacity leads to the continuous increase of social
costs, but the impact on coalition profits are not monotonous.
Although reserve capacity will increase social cost, the social
cost of centralized market mechanism is lower than that of
deregulated market mechanism. The coalition profits without
financial incentive under the centralized market mechanism is
14008
FIGURE 11. The impact of reserve capacity on coalition profits and social
costs under the different market mechanism.
zero (blue bar is invisible), while the coalition profits under
the deregulated market mechanism is positive.
For the numerical results from the deregulated market
mechanism, from figure 7 and figure 9, both PTC and ITC
financial incentives can increase the installed capacity of WT
and ESS to improve the utilization of wind energy. From
figure 8 and figure 10, both PTC and ITC financial incen-
tives can increase the coalition profit, but the effect of these
two financial incentives in reducing social costs is different.
Besides, the revenue TIP of WT and ESS from the energy
and auxiliary service market under the deregulation market
mechanism is positive.
The impact of two financial incentives (PTC and ITC)
on investors and power system economy under two mar-
ket mechanisms (centralized and deregulated market mech-
anism) has been studied based on modified IEEE 6-bus test
system. The average computation time for each case under the
centralized and deregulated market mechanism is 3 seconds
and 60 seconds respectively. The former is to maximize social
welfare problem, which is a tractable LP, while the latter is to
maximizing investment profit problem, which is a complex
MILP.
B. MODIFIED IEEE 30-BUS TEST SYSTEM
In this section, we apply the proposed model and the solu-
tion methodology considering the modified IEEE 30-bus test
system to illustrate the fragilities of financial incentive. The
modified test system has 30 nodes that are connected through
41 transmission lines. The set of buses where WT can be
installed includes buses 2, 6, 10, 12, and the set of buses
where ESS can be installed includes buses 4, 10, 15, 27.
The maximum load demand in the modified IEEE 30-bus test
system are 125% of their original values, and the capacity of
lines is 90% of their original values. The parameter of TG are
given in Table 3.
In this test system, the difference between the production
cost of TG and the capital cost of WT is relatively small.
VOLUME 8, 2020
K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms
TABLE 3. Thermal generator data modified IEEE 30-bus test system.
Besides, we focus on the comparison of electrical informa-
tion and economic indicators result from centralized and
deregulated market mechanisms base on the same financial
incentives.
To begin with, the installed capacity of WT and ESS as
well as the amount of wind energy utilization and curtailment
result from under the different market mechanism based on
PTC financial subsidies are given in Figure 12.
FIGURE 12. Electric information indicators comparison based on PTC
financial subsidy under centralized and deregulated market mechanism.
Figure 12 indicates that for the installed capacity of WT
and ESS, the result of centralized market mechanism is
greater than that of deregulated market mechanism. Besides,
the amount of wind energy utilization and curtailment under
the centralized market mechanism is also greater than that
deregulated market mechanism. Meanwhile, the comparison
results of economic indicators are shown in Figure 13.
Under the same PTC financial subsidy intensity, the cen-
tralized market mechanism has a lower social cost, while
deregulation market mechanism has a higher coalition profit.
This result is consistent with Figures 8 and 10. We use the
LMP to explain the huge differences between the electrical
information and economic indicators under the two market
mechanisms. For example, without financial subsidies, the
LMP of each node under the two market mechanisms is given
in Figure 14.
The average value, minimum value and maximum value of
LMP under the centralized market mechanism are 31 $/MWh,
0 $/MWh, and 258 $/MWh, respectively. The average value,
VOLUME 8, 2020
FIGURE 13. Economic indicators comparison based on PTC financial
subsidy under centralized and deregulated market mechanism.
FIGURE 14. Locational marginal price of each node under centralized and
deregulated market mechanism without financial subsidy.
minimum value and maximum value of LMP under the
deregulated market mechanism are 77 $/MWh, 57 $/MWh,
and 1156 $/MWh, respectively. Coalition under deregulated
market mechanism tend to lower the installed capacity of
WT and ESS so that the increase of LMP, which will in turn
bring higher profits to coalition. Besides, it should be noted
that there would be zero marginal cost under the centralized
market mechanism (from 1:00 to 10:00).
Besides, the installed capacity of WT and ESS as well as
the amount of wind energy utilization and curtailment result
from under the different market mechanism based on ITC
financial subsidies are given in Figure 15.
As mentioned above, the installed capacity of WT and ESS
under the centralized market mechanism is larger than that of
deregulated market. However, when the proportion of ITC
financial incentive subsidy reaches 30%, the amount of wind
energy curtailment is far greater than the amount of wind
energy utilization, which shows that the intensity of financial
subsidy is unreasonable. In the PTC financial incentive in
Figure 12, there are also wind energy curtailment far greater
than wind energy utilization. Although the two financial
14009
 K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms
FIGURE 15. Electric information indicators comparison based on ITC
financial subsidy under centralized and deregulated market mechanism.
incentives are the same, the phenomenon does not exist in the
IEEE 6-bus system. Therefore, when the difference between
TG production cost and WT capital cost is small, financial
incentive policies should be carefully designed. Furthermore,
the results of social cost and coalition profit based on ITC
financial subsidy under the two market mechanisms are
shown in Figure 16.
FIGURE 16. Economic indicators comparison based on ITC financial
subsidy under centralized and deregulated market mechanism.
We note that ITC financial subsidies can effectively reduce
social costs under both centralized and deregulated market
mechanisms. Besides, the coalition profit under the cen-
tralized market mechanism completely covers ITC financial
subsidies.
Based on the modified IEEE 30-bus test system, the elec-
trical information and economic indicators under the same
financial incentive are analyzed. When the difference
between thermal power production cost and capital cost
is small, the driving force of financial incentive policy is
not significant. Besides, the incentive effect of PTC pol-
icy under deregulated market mechanism is not significant
14010
(Figure 7 and Figure 12), while that of ITC policy under both
market mechanisms is significant. The average computation
time for each case under the centralized and deregulated
market mechanism is 5 s and 400 s respectively. Large-scale
system will increase the computation time of coordinated
planning problem under deregulated market mechanism,
which is acceptable for offline simulation.
In summary, under the centralized market mechanism,
the total coalition profit under the financial incentive of PTC
and ITC does not exceed the financial subsidy. Specifically,
the total coalition profits (TCPPTC ) under PTC financial
incentive is lower than financial subsidy, while the total coali-
tion profits (TCPITC ) under ITC financial incentive is equal
to financial subsidy. Under deregulated market mechanism,
the total coalition profit under the financial incentive of PTC
and ITC is no less than the financial subsidy.
VII. CONCLUSION
This paper presents a coordinated planning and operation
model of power system considering market mechanism and
financial incentive policy. The results from the presented case
study yield the following conclusions:
For the centralized market mechanism, the revenue of
coalition from energy and auxiliary market are non-positive
(TIP is non-positive). For the deregulated market mechanism,
the revenue of coalition from energy and auxiliary market are
positive (TIP is positive).
The incentive effect of PTC policy under the centralized
market mechanism is significant, which can increase the
installed capacity of WT and ESS, and reduce the social cost.
However, PTC policy under deregulated market mechanism
is fragile. The incentive effect of ITC policy under centralized
and deregulated market mechanism is significant, which can
increase the installed capacity of WT and ESS, and reduce
social costs.
The installed capacity of WT and ESS under the central-
ized market mechanism is larger than that under the dereg-
ulated market mechanism, as well as the amount of wind
energy utilization and curtailment. The social cost under the
centralized market mechanism is lower than that under the
deregulated market mechanism, while the coalition profit
under the centralized market mechanism is lower than that
under the deregulated market mechanism. This phenomenon
can be explained by the LMP. Compared with the deregulated
market mechanism, the LMP under the centralized market
mechanism is very small (part of LMP is zero or negative),
and the difference of nodal price in temporal and spatial is
not significant.
Future research will focus on the design of financial
incentives and the impact of uncertainty. Design reasonable
financial incentives to avoid transitional investment in RES.
Uncertainty will affect the return of investors, then it is
of great significance to study the investment of renewable
energy by applying risk management theory.
VOLUME 8, 2020
K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms

APPENDIX λtes,d,i ptes,d,i = 0 (90)

KKT CONDITIONS FOR MAXIMIZING SOCIAL WELFARE λ̄tes,d,i
ptes,d,i − sces nes,i = 0


(91)
KKT condition includes three parts: primal constraint, first-
λtes,ur,iptes,d,i /ηd + ures,i
t


order optimality condition and complementary slackness =0 (92)
λ̄tes,ur,i
pes,d,i /ηd + ures,i − sces nes,i = 0
t t


constraints. Note that the Lagrange function is omitted. (93)
Firstly, the primal constraint consists of equation (1)-(4), λtes,dr,iηc ptes,c,i + dres,i
t


=0 (94)
(6)-(7), and (9)-(17), and (19)-(22).
λ̄tes,dr,i
ηc pes,c,i + dres,i − sces nes,i = 0
t t


(95)
Secondly, the first-order optimality condition are as
t t
!
follows. X
τ
X
τ
λes,r,i ures,i − ees,i −
t t 0
ηc pes,c,i + pes,d,i /ηd =0
bi − λtn,i − λtg,i + λ̄tg,i = 0 (68) τ =1 τ =1
(96)
−λn,i − λw,i + λ̄w,i = 0
t t t
(69) t t
ces,d /ηd −λn,i −λes,d,i + λ̄es,d,i + λ̄es,ur,i − λes,ur,i /ηd ηc pτes,c,i + pτes,d,i /ηd
t t t t t
X X
λ̄tes,r,i e0es,i +

X T τ =1 τ =1
λtes,r,i − λ̄tes,r,i /ηd +λtes,s,i /ηd = 0 − ssoc t



+ (70) es nes,i + dres,i =0 (97)
τ =t λn,es,i nes,i = 0 (98)
ces,c ηc + λtn,i −λtes,c,i + λ̄tes,c,i + ηc λ̄tes,dr,i − λtes,dr,i

n  o
λtij −pmaxij − Bij θ t
i − θ t
j =0 (99)
X T
λ̄tes,r,i − λtes,r,i − ηc λtes,s,i = 0

n   o
+ηc (71) λ̄tij Bij θit − θjt − pmax =0 (100)
ij
τ =t X X X
cug,i − λtg,ur,i + λ̄tg,ur,i + λ̄tg,i − λtur = 0 (72) λtur rd ptd,i − λtur t
urg,i − λtur t
ures,i =0 (101)
i∈Nd i∈Ng i∈Nes
cg,i − λg,dr,i + λ̄g,dr,i + λg,i − λdr = 0
d t t t t
(73) X X X
ces,i − λes,ur,i + λ̄es,ur,i + λes,r,i − λur = 0
u t t t t
(74) λtdr rd ptd,i − λtdr t
drg,i − λtdr t
dres,i = 0. (102)
i∈Nd i∈Ng i∈Nes
ces,i − λes,dr,i + λ̄es,dr,i + λ̄es,r,i − λdr = 0
d t t t t
(75)
X T REFERENCES
cbw − λn,w,i − scw λ̄tw,i = 0 (76) [1] D. J. Hess and M. Renner, ‘‘Conservative political parties and energy
t=1 transitions in Europe: Opposition to climate mitigation policies,’’ Renew.
T Sustain. Energy Rev., vol. 104, pp. 419–428, Apr. 2019.
X [2] G. Wang, Q. Zhang, Y. Li, and B. C. Mclellan, ‘‘Efficient and equi-
cbes − λn,es,i − sces λ̄tes,c,i + λ̄tes,d,i table allocation of renewable portfolio standards targets among China’s
t=1 provinces,’’ Energy Policy, vol. 125, pp. 170–180, Feb. 2019.
T [3] S. Saint Akadiri, A. A. Alola, A. C. Akadiri, and U. V. Alola, ‘‘Renewable
X energy consumption in EU-28 countries: Policy toward pollution mitiga-
+ λ̄tes,ur,i + λ̄tes,dr,i − ssoc λ̄tes,r,i = 0


es (77) tion and economic sustainability,’’ Energy Policy, vol. 132, pp. 803–810,
t=1 Sep. 2019.
[4] N. A. Obeng-Darko, ‘‘Why ghana will not achieve its renewable energy
X   X  
Bij λtn,i −λtn,j + Bij λ̄tij −λtij +λtji − λ̄tji = 0 (78) target for electricity. Policy, legal and regulatory implications,’’ Energy
j∈N j∈N Policy, vol. 128, pp. 75–83, May 2019.
[5] M. Maulidia, P. Dargusch, P. Ashworth, and F. Ardiansyah, ‘‘Rethink-
Thirdly, the complementary slackness constraints are as ing renewable energy targets and electricity sector reform in Indonesia:
follows. A private sector perspective,’’ Renew. Sustain. Energy Rev., vol. 101,
pp. 231–247, Mar. 2019.
[6] G. Li, G. Li, and M. Zhou, ‘‘Comprehensive evaluation model of wind
λtg,ur,i urg,i
t
=0 (79) power accommodation ability based on macroscopic and microscopic
 
indicators,’’ Protection Control Mod. Power Syst., vol. 4, no. 1, p. 19, 2019.
λ̄tg,ur,i urg,i
t max
− urg,i =0 (80) [7] H. Hamidpour, J. Aghaei, S. Pirouzi, S. Dehghan, and T. Niknam, ‘‘Flex-
ible, reliable, and renewable power system resource expansion planning
λtg,dr,i drg,i
t
=0 (81) considering energy storage systems and demand response programs,’’ IET
Renew. Power Gener., vol. 13, no. 11, pp. 1862–1872, Aug. 2019.
 
λ̄tg,dr,i drg,i
t max
− drg,i =0 (82) [8] A. Ehsan and Q. Yang, ‘‘Coordinated investment planning of distributed
  multi-type stochastic generation and battery storage in active distribution
λtg,i pmin
g,i + dr t
g,i − p t
g,i = 0 (83) networks,’’ IEEE Trans. Sustain. Energy, vol. 10, no. 4, pp. 1813–1822,
  Oct. 2019.
λ̄tg,i ptg,i − pmax t
g,i + urg,i = 0 (84) [9] A. Moreira, D. Pozo, A. Street, and E. Sauma, ‘‘Reliable renewable
generation and transmission expansion planning: Co-optimizing system’s
λtw,i ptw,i = 0 (85) resources for meeting renewable targets,’’ IEEE Trans. Power Syst., vol. 32,
no. 4, pp. 3246–3257, Jul. 2017.
λ̄tw,i ptw,i − scw nw,i


=0 (86) [10] Q. Huang, Y. Xu, T. Wang, and C. A. Courcoubetis, ‘‘Market mechanisms
for cooperative operation of price-maker energy storage in a power net-
λn,w,i nw,i = 0 (87) work,’’ IEEE Trans. Power Syst., vol. 33, no. 3, pp. 3013–3028, May 2018.
λtes,c,i ptes,c,i = 0 (88) [11] R. Singh, P. R. Kumar, and L. Xie, ‘‘Decentralized control via dynamic
stochastic prices: The independent system operator problem,’’ IEEE Trans.
λ̄tes,c,i ptes,c,i − sces nes,i = 0


(89) Autom. Control, vol. 63, no. 10, pp. 3206–3220, Oct. 2018.

VOLUME 8, 2020 14011

 K. Tian et al.: Joint Planning and Operation for Renewable-Storage Under Different Financial Incentives and Market Mechanisms
[12] S. Agrali, F. Terzi, E. Canakoglu, E. Adiyeke, and Y. Arikan, ‘‘Energy
investment planning at a private company: A mathematical programming-
based model and its application in Turkey,’’ IEEE Trans. Power Syst.,
vol. 32, no. 6, pp. 4180–4187, Nov. 2017.
[13] J. Valinejad, T. Barforoshi, M. Marzband, E. Pouresmaeil, R. Godina, and
J. P. S. Català, ‘‘Investment incentives in competitive electricity markets,’’
Appl. Sci., vol. 8, no. 10, p. 1978, Oct. 2018.
[14] J. Valinejad, M. Marzband, M. Elsdon, A. Saad Al-Sumaiti, and
T. Barforoushi, ‘‘Dynamic carbon-constrained EPEC model for strategic
generation investment incentives with the aim of reducing CO2 emissions,’’
Energies, vol. 12, no. 24, p. 4813, Dec. 2019.
[15] F. R. Badal, P. Das, S. K. Sarker, and S. K. Das, ‘‘A survey on control issues
in renewable energy integration and microgrid,’’ Protection Control Mod.
Power Syst., vol. 4, no. 1, p. 8, 2019.
[16] G. Aquila, E. D. O. Pamplona, A. R. D. Queiroz, P. R. Junior, and
M. N. Fonseca, ‘‘An overview of incentive policies for the expansion of
renewable energy generation in electricity power systems and the Brazil-
ian experience,’’ Renew. Sustain. Energy Rev., vol. 70, pp. 1090–1098,
Apr. 2017.
[17] S. Pineda, T. K. Boomsma, and S. Wogrin, ‘‘Renewable generation expan-
sion under different support schemes: A stochastic equilibrium approach,’’
Eur. J. Oper. Res., vol. 266, no. 3, pp. 1086–1099, May 2018.
[18] X. Yang, L. He, Y. Xia, and Y. Chen, ‘‘Effect of government subsidies
on renewable energy investments: The threshold effect,’’ Energy Policy,
vol. 132, pp. 156–166, Sep. 2019.
[19] F. Nicolli and F. Vona, ‘‘Energy market liberalization and renewable
energy policies in OECD countries,’’ Energy Policy, vol. 128, pp. 853–867,
May 2019.
[20] G. Doorman and A. Botterud, ‘‘Analysis of generation investment under
different market designs,’’ IEEE Trans. Power Syst., vol. 23, no. 3,
pp. 859–867, Aug. 2008.
[21] J. Valinejad, M. Marzband, Y. Xu, H. Uppal, A. Saad Al-Sumaiti, and
T. Barforoshi, ‘‘Dynamic behavior of multi-carrier energy market in view
of investment incentives,’’ Elect. Eng., vol. 101, no. 3, pp. 1033–1051,
Sep. 2019.
[22] G. Capizzi, G. Lo Sciuto, C. Napoli, and E. Tramontana, ‘‘Advanced and
adaptive dispatch for smart grids by means of predictive models,’’ IEEE
Trans. Smart Grid, vol. 9, no. 6, pp. 6684–6691, Nov. 2018.
[23] S. Dehghan and N. Amjady, ‘‘Robust transmission and energy storage
expansion planning in wind farm-integrated power systems considering
transmission switching,’’ IEEE Trans. Sustain. Energy, vol. 7, no. 2,
pp. 765–774, Apr. 2016.
[24] N. G. Cobos, J. M. Arroyo, N. Alguacil, and A. Street, ‘‘Robust energy
and reserve scheduling under wind uncertainty considering fast-acting
generators,’’ IEEE Trans. Sustain. Energy, vol. 10, no. 4, pp. 2142–2151,
Oct. 2019.
[25] J. Yang, J. Zhao, J. Qiu, and F. Wen, ‘‘A distribution market clearing
mechanism for renewable generation units with zero marginal costs,’’ IEEE
Trans. Ind. Informat., vol. 15, no. 8, pp. 4775–4787, Aug. 2019.
[26] B. Xu, Y. Wang, Y. Dvorkin, R. Fernandez-Blanco, C. A. Silva-Monroy,
J.-P. Watson, and D. S. Kirschen, ‘‘Scalable planning for energy storage
in energy and reserve markets,’’ IEEE Trans. Power Syst., vol. 32, no. 6,
pp. 4515–4527, Nov. 2017.
[27] Z. Shi, H. Liang, S. Huang, and V. Dinavahi, ‘‘Distributionally robust
chance-constrained energy management for islanded microgrids,’’ IEEE
Trans. Smart Grid, vol. 10, no. 2, pp. 2234–2244, Mar. 2019.
[28] B. V. Solanki, A. Raghurajan, K. Bhattacharya, and C. A. Cañizares,
‘‘Including smart loads for optimal demand response in integrated energy
management systems for isolated microgridss,’’ IEEE Trans. Smart Grid.,
vol. 8, no. 4, pp. 1739–1748, Jul. 2017.
[29] Z. Li, Q. Guo, H. Sun, and J. Wang, ‘‘Sufficient conditions for exact relax-
ation of complementarity constraints for storage-concerned economic dis-
patch,’’ IEEE Trans. Power Syst., vol. 31, no. 2, pp. 1653–1654, Mar. 2016.
[30] F. Arasteh and G. H. Riahy, ‘‘Social welfare maximisation of market based
wind integrated power systems by simultaneous coordination of transmis-
sion switching and demand response programs,’’ IET Renew. Power Gener.,
vol. 13, no. 7, pp. 1037–1049, May 2019.
[31] B. Stott, J. Jardim, and O. Alsac, ‘‘DC power flow revisited,’’ IEEE Trans.
Power Syst., vol. 24, no. 3, pp. 1290–1300, Aug. 2009.
[32] J. A. Taylor, ‘‘Financial storage rights,’’ IEEE Trans. Power Syst., vol. 30,
no. 2, pp. 997–1005, Mar. 2015.
[33] Q. Huang, Y. Xu, and C. Courcoubetis, ‘‘Financial incentives for joint
storage planning and operation in energy and regulation markets,’’ IEEE
Trans. Power Syst., vol. 34, no. 5, pp. 3326–3339, Sep. 2019.
14012
[34] M. Li, ‘‘Generalized Lagrange multiplier method and KKT conditions with
an application to distributed optimization,’’ IEEE Trans. Circuits Syst. II,
Exp. Briefs, vol. 66, no. 2, pp. 252–256, Feb. 2019.
[35] E. Nasrolahpour, S. J. Kazempour, H. Zareipour, and W. D. Rosehart,
‘‘Strategic sizing of energy storage facilities in electricity markets,’’ IEEE
Trans. Sustain. Energy, vol. 7, no. 4, pp. 1462–1472, Oct. 2016.
[36] A. Sinha, P. Malo, and K. Deb, ‘‘A review on bilevel optimization: From
classical to evolutionary approaches and applications,’’ IEEE Trans. Evol.
Comput., vol. 22, no. 2, pp. 276–295, Apr. 2018.
[37] H. Mohsenian-Rad, ‘‘Coordinated price-maker operation of large energy
storage units in nodal energy markets,’’ IEEE Trans. Power Syst., vol. 31,
no. 1, pp. 786–797, Jan. 2016.
[38] K. Pandžić, H. Pandžić, and I. Kuzle, ‘‘Coordination of regulated and mer-
chant energy storage investments,’’ IEEE Trans. Sustain. Energy, vol. 9,
no. 3, pp. 1244–1254, Jul. 2018.
[39] S. Siddiqui, S. A. Gabriel, ‘‘An SOS1-based approach for solving MPECs
with a natural gas market application,’’ Netw. Spatial Econ., vol. 13,
pp. 205–227, Jul. 2013.
[40] L. Baringo and A. Baringo, ‘‘A stochastic adaptive robust optimization
approach for the generation and transmission expansion planning,’’ IEEE
Trans. Power Syst., vol. 33, no. 1, pp. 792–802, Jan. 2018.
[41] G. R. Timilsina, L. Kurdgelashvili, and P. A. Narbel, ‘‘Solar energy:
Markets, economics and policies,’’ Renew. Sustain. Energy Rev., vol. 16,
no. 1, pp. 449–465, Jan. 2012.
[42] M. B. Mcelroy and X. Chen, ‘‘Wind and solar power in the United States:
Status and prospects,’’ CSEE. J. Power Energy Syst., vol. 3, no. 1, pp. 1–6,
Mar. 2017.
[43] R. D. Zimmerman, C. E. Murillo-Sánchez, and D. Gan, ‘‘MAT-
POWER: A MATLAB power system simulation package,’’ Manual, Power
Syst. Eng. Res. Center, Ithaca, NY, USA, 1997. [Online]. Available:
http://www.pserc.cornell.edu/matpower/manual.pdfs
KUNPENG TIAN received the M.S. degree in
electrical engineering from the University of
Shanghai for Science and Technology, Shanghai,
China, in 2018, where he is currently pursuing the
Ph.D. degree. His research interests include power
system operation and investment planning.
WEIQING SUN received the B.S., M.S., and
Ph.D. degrees in electrical engineering from
Shanghai Jiao tong University, Shanghai, China,
in 2007, 2009, and 2013, respectively. He is cur-
rently an Associate Professor with the University
of Shanghai for Science and Technology, Shang-
hai. His current research interests are renewable
energy, power system planning, and optimization
theory.
DONG HAN received the Ph.D. degree in elec-
trical engineering from Shanghai Jiao tong Uni-
versity, Shanghai, China, in 2016. He is currently
a Lecturer with the University of Shanghai for
Science and Technology, Shanghai. His current
research interests are electricity market and opti-
mization theory.
CE YANG received the B.S. degree in electri-
cal engineering and automation from Shanghai
Dianji University, Shanghai, China, in 2014. He is
currently pursuing the Ph.D. degree with the
Department of Control Science and Engineering,
University of Shanghai for Science and Tech-
nology, Shanghai. His research interests include
power system automation and energy management
systems.
VOLUME 8, 2020