This article has been accepted for publication in a future issue of this journal, but has not been

edited. Content will change prior to final publication.

JOURNAL OF MODERN POWER SYSTEMS AND CLEAN ENERGY, VOL. XX, NO. XX, XXXX 1

Exploiting Flexibility of Integrated Demand

Response to Alleviate Power Flow Violation During
Line Tripping Contingency
Tong Cheng, Student Member, IEEE, Zhenfei Tan, Student Member, IEEE and Haiwang Zhong, Senior Member,
IEEE

gc
Abstract—Multiple energy integrations provide great oppor- Um Set of gas compressors at gas node m
tunities for economical and efficient resource utilization. In U gl Set of contracted gas loads
the meantime, power system operation requires enough flexible Umgl
Set of gas loads at gas node m
resources to deal with contingencies such as transmission line
tripping. Besides economic benefits, this paper focuses on the U gs Set of gas suppliers
gs
security benefits multiple energy integration can provide. This Um Set of gas suppliers at gas node m
paper first proposes an operation scheme to coordinate multiple LP Set of power transmission lines
energy production and local system consumption considering LH Set of pipes in heat networks
transmission networks. The integrated flexibility model, con- H,+/−
Lq Set of pipes with fluid flowing into/out of node q
structed by the feasible region of integrated demand response
(IDR), is then formulated to aggregate and describe local flexi- LH,s/b Set of pipes in heat supply/return networks
bility. Combined with system security constraints, a multi-energy LHs/Hl Set of pipes connected with heat source/load
system operation model is formulated to schedule multiple energy LHq Set of pipes connected with node q
production, transmission and consumption. The effects of local g Index for generator units
system flexibility on alleviating power flow violations during
i Index for units
N-1 line tripping contingencies are then analyzed through a
multi-energy system case. The results show that local system j Index for gas compressors
flexibility can not only reduce the system’s operating costs, but k Index for power transmission lines
also reduce the probability of power flow congestion or violations l Index for gas loads
by approximately 68.8% during N-1 line tripping contingencies. m, n, q Index for nodes
p Index for heat pipes
Index Terms—Multi-energy system, integrated flexibility, fea- s Index for gas suppliers
sible region, integrated demand response, N-1 security. t Index for time intervals
Parameters and Constants
N OMENCLATURE
Indices and Sets σ A fixed penalty factor
Fk,max The capacity of transmission line k
NP Set of power system nodes
GSDFk,n The generation shift distribution factor of
NG Set of gas system nodes
node n to line k
NH Set of heat system nodes
gn N (g) The node of generator g
Nm Set of gas nodes connected with gas node m
Pg,max /Pg,min The maximum/minimum capacity of gener-
T Set of time intervals
ator g
UG Set of generators
Rampg The ramp rate of generator g
U CF Set of coal-fired generators
SRtr The required spinning reserve of the system
U GF Set of gas-fired generators
at time t
U CHP Set of combined heat and power units
tof f
The minimum shutdown time of generator g
U GB Set of gas boilers g,min
ton The minimum start-up time of generator g
U gc Set of gas compressors g,min
_____________________________________ Cmn The gas pipeline constant from node m to n
Manuscript received: August 3, 2021; revised: November 13, 2021; ac- ρi The gas contract price for gas consumer i
cepted: January 25, 2022. Date of CrossCheck: January 25, 2022. Date of aj , bj , cj Gas consumption-related constants of com-
online publication: XX XX, XXXX.
This work was supported by State Grid Corporation of China project pressor j
‘Research on multi-energy system energy conversion simulation and energy Hj,max /Hj,min The maximum/minimum horsepower of gas
efficiency evaluation’(SGTYHT/18-JS-206). compressor j
This article is distributed under the terms of the Creative Commons Attri-
bution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). πm,max /πm,minThe maximum/minimum pressure at gas
T. Cheng, Z. Tan, and H. Zhong (corresponding author) are with the State node m
Key Laboratory of Power Systems, the Department of Electrical Engineering, vs,max /vs,min The maximum/minimum gas injection of gas
Tsinghua University, Beijing, China (e-mail: ct20@mails.tsinghua.edu.cn;
zftan@outlook.com; zhonghw@tsinghua.edu.cn). supplier s
DOI: 10.35833/MPCE.2021.000535 kj1 , kj2 , α Gas flow-related constants of compressor j
 This article has been accepted for publication in a future issue of this journal, but has not been edited. Content will change prior to final publication.
2 JOURNAL OF MODERN POWER SYSTEMS AND CLEAN ENERGY, VOL. XX, NO. XX, XXXX

λ Heat conduction coefficient situations. Taking the power system as an example, the flexi-
c The specific heat of the fluid bility may help to avoid power flow violations and thus reduce
Ciboiler The gas-heat conversion coefficient of boiler transmission line investment, which will be further discussed
i in this paper. Actually, security is always a key requirement of
CiCHP The heat-power ratio of CHP i electric power system operation [1]. The security-constrained
Lp The length of pipe p unit commitment (SCUC) problem, which considers both the
mp The fluid flow of pipe p normal state and the N-1 contingency state, is commonly used
s/b
Tmin / max The minimum/maximum temperature of the to decide the short-term schedule of power systems.
fluid supply/return pipe The power system unit commitment problem considering
Ta The environment temperature N-1 contingencies can be divided into two categories: preven-
Variables tive control and corrective control. For both control methods,
the system operator considers all possible N-1 conditions and
Pgt The power output of generator g at time t derives a day-ahead schedule result. The difference lies in
Ugt The working status of generator g at time t the fact that preventive control requires the system to oper-
Ygt The start-up status of generator g at time t ate safely without changing the generators’ schedule, while
Zgt The shut-down status of generator g at time t corrective control allows generators to change their output to
Pkt The power transmission of line k at time t relieve the power flow violation in a given time.
L
Pnt The power load of node n at time t Reference [2] proposes a DC optimal power flow (DCOPF)-
slk Slack variables introduced for line k based preventive SCUC model with N-1 reliability, in which
πm The gas pressure of node m all contingency conditions are embedded into the optimiza-
Ff,j The gas consumption of compressor j tion problem through integer variables. Furthermore, a line
fmn The gas flow from gas node m to n outage distribution factor-based method is proposed in [3] to
Hj The horsepower of gas compressor j reduce the computational burden of the problem. However,
Lit The flow of the gas input of unit i at time t as stated in [4], the preventive control strategy does not
Ll The flow of gas load l consider the generators’ real-time adjustment ability, thus
vs The flow of gas supply s making the dispatch order too conservative and compromising
s/l
Qq The heat supply/load of node q the system’s economic efficiency. A DCOPF-based corrective
Qit The heat output of unit i at time t SCUC model considering the long- and short-term emergency
s/b
Tp,in Temperature of the inflow of supply/return pipe (LTE/STE) rates is then proposed. The application of the
p so-called LTE/STE allows the temporal exceedance of trans-
s/b
Tp,out Temperature of the outflow of supply/return pipe mission capacity in post-contingency operation to make the
p system schedule less conservative according to [5]. An AC
s/b
Tq Temperature of the outflow of node q in the contingency dispatch model based on preventive/corrective
supply/return system control is proposed in [6] to balance the system’s economic
Wit The gas cost of unit i at time t and security properties.
However, few studies realize the potential of demand-side
I. I NTRODUCTION adjustment on N-1 reliability, which may lead to unnecessary
investment in new transmission lines. Demand response pro-
ITH the concept of carbon-free energy system transi-
W tion, multi-energy systems have gained much attention
due to their superiority in utilizing complementary energy
grams have long been used to enhance system economic and
security performance, such as to maximize social welfare [7]
and avoid voltage collapse [8]. With multi-energy interaction,
resources and improving energy efficiency. conventional demand response programs can also turn into
Large-scale interconnected energy systems provide great integrated demand response (IDR) programs [9]. In this way,
opportunities for economical and efficient resource utilization the demand side can also change its load composition to
in a larger spatial range. In the on-going practice of multi- adjust the line flow taking advantage of the flexibility of
energy integration, different energy carriers are coupled and other systems. For example, the local system can use more
integrated in various sectors of the entire energy supply heat instead of electricity to alleviate congestion in power
chain, from energy production, transmission/transportation, to transmission lines.
distribution and consumption. Multi-energy carriers interact In fact, the integrated demand response (IDR) has been
with each other at various spatial levels, from regional systems widely studied in recent years. IDR programs can be utilized
(e.g., urban energy supply infrastructure), downscaling to to inhibit demand, adjust load curves and improve customer
local systems (e.g., smart buildings, energy communities, and satisfaction through different price signals and operation strate-
industrial parks). These interactions not only provide chances gies [10]. An incentive-based IDR program and its model
for system operators to maximize social welfare but also play are proposed in [11] to reduce the total cost of the multi-
an important role in enhancing the resilience and stability of energy aggregator. In [12], a price-based IDR scheme is
the whole system. proposed for integrated electricity and natural gas systems
More specifically, the flexibility provided by multi-energy to demonstrate its potential of switching energy resources to
integration may help any single system handle contingency maximize profits and a potential game model is proposed.
 This article has been accepted for publication in a future issue of this journal, but has not been edited. Content will change prior to final publication.
TONG et al.: EXPLOITING FLEXIBILITY OF INTEGRATED DEMAND RESPONSE TO ALLEVIATE POWER FLOW VIOLATION DURING LINE TRIPPING...
In [13], a demand response program of smart buildings in
integrated heat and electricity systemd is studied to provide
heat and electricity balancing power. A production schedul-
ing model for manufacturers considering electricity and gas
demand response is established in [14] to save energy costs.
In [15], an IDR optimization model that considers network
constraints is proposed.
Despite the research progress in IDR models like price-
based, game theory-based and smart energy hub (SEH) models
and the coordinated optimization scheme of IDR with net-
works, few studies focus on and quantify IDR’s impact on
the security margin of the system. In [16], an N-1 security-
constrained scheduling model for integrated electricity and
gas systems is proposed, through which the influence of
electricity transmission lines and gas pipeline tripping on
the integrated system is further discussed. In [17], the static
security influence of the exit of coupling elements in inte-
grated energy systems is analyzed. However, the potential of
integrated energy flexibility in dealing with contingencies is
not considered.
In mainstream IDR research, IDR usually operates in a cer-
tain status according to given price signals, certain incentives
or a game model. However, to perform quantitative research on
the impact of IDR on the system’s security level, the feasible
region of the local integrated energy should be determined
and then combined with the network constraints to form a
coordinated optimization problem. The feasible region of IDR,
called the integrated flexibility region, can be established to
quantify the flexibility the local IDR can provide based on
our previous work [18]. Then, we apply a corrective N-1
contingency model and loosen the line capacity constraints.
By introducing slack variables to these constraints, the benefits
of the IDR’s flexibility in alleviating violations are further
reflected.
The main contributions of this paper include the following
two aspects.
1) A coordination scheme is proposed to utilize the flexibil-
ity of multi-energy conversion to alleviate power system N-1
contingency violations.
2) An explicit model is formulated to characterize the
feasible region of the IDR, which can be embedded into the
power system N-1 schedule problem without specific local
system information, and thus the effect of the IDR’s flexibility
is analyzed.
The rest of this paper is organized as follows. In Section II,
the system framework, including its structure and coordination
scheme, is stated. In Section III, the specific optimization
model, including IDR’s feasible region model and network
constraints model, is formulated. In Section IV, the perfor-
mance of the proposed framework is demonstrated through a
multi-energy case system. Conclusions are drawn in Section
V.
II. IDR F LEXIBILITY AND C OORDINATION S CHEME
A. System Structure
A conceptual framework of the regional and local multi-
energy systems is illustrated in Fig. 1. The regional system
3
consists of multiple energy resources, energy converters, and
energy networks. In this study, the energy resources are the
only interface through which the regional system imports
energy from external systems. Energy resources include, for
instance, natural gas input, electricity from the external grid,
and fossil fuels to drive generators. The overall operation
costs of the entire system are only associated with the price
and consumed amount of energy resources. Energy converters
enable theintegration of multi-energy flows by converting a
single type of energy input to other kinds of energy outputs.
The networks connect energy resources, converters, and loads.
In the existing regulation framework, networks are usually mo-
nopolized by a utility for a certain region. The regional system
operator (RSO) optimally schedules the energy converters to
supply the multi-energy demand (MED) of consumers with
minimal operation cost.
(OHFWULFLW\
'LUHFWO\
&RDO *DV VXSSOLHG
+HDW 0('
*DV &HQWUDOL]HGJHQHUDWLRQ
:LQG DQG GHOLYHU\
0('RI
6(+
6RODU ORFDO6(+
3ULPDU\(QHUJ\
8WLOLW\6LGH 3'3 'HPDQG6LGH
Fig. 1. System framework.
The energy consumers connected to the regional system
can be classified into two types, namely the directly supplied
loads and local multi-energy systems. The former is directly
connected to a node of the centralized regional network, and
its demand turns out to be the fixed boundary condition in
utility-level system scheduling. The latter is equipped with
local energy converters, which convert the energy supplied
by the regional network to serve the terminal multiple energy
demand. With energy converters, the local multi-energy system
is endowed with the ability to adjust its energy inputs without
affecting the terminal load demand, which provides operation
flexibility to the regional system. This feature of the local
multi-energy system has been verified in [19], where local
systems are modelled as energy hubs, and an integrated model
is established to optimally schedule the centralized and local
systems. In real-world practice, however, local systems and
regional networks are owned and operated by different entities.
In general, the RSO can neither control nor obtain information
access to consumer-owned devices. Hence, it is impractical
to integrate the flexibility of local IDR through centralized
optimization. As demonstrated in Fig. 1, local systems are in-
terfaced with the regional system at the property division point
(PDP). A viable method is to obtain an external equivalence of
each local system and schedule the regional system with an
external description. The external characteristics of the IDR
of the local multi-energy system are modelled and described
explicitly via a novel concept, namely integrated flexibility.
 This article has been accepted for publication in a future issue of this journal, but has not been edited. Content will change prior to final publication.
4 JOURNAL OF MODERN POWER SYSTEMS AND CLEAN ENERGY, VOL. XX, NO. XX, XXXX

B. Integrated Energy Flexibility
The integrated flexibility is defined as the ability of the local
multi-energy system equipped with energy converters to serve
its fixed terminal multi-energy loads with adjustable energy
inputs through IDR programs. This ability naturally arises
from the multi-energy synergy at the local level. The local
where the multi-row equalities g(·) ≤ 0 represent the operat-
ing constraints of the energy hub. A Detailed formulation will
be derived in Section III.
Mathematically, the IDR feasible region is then formulated
as the projection of the operating feasible region Φi (D ls,i )
onto the subspace of the input vector space, i.e.,
ls,i
∈ Φi (D ls,i ),
  
system ‘reprocesses’ the multi-energy flow imported from the i ls,i ls,i  ∃ V

Ω (D ) = Vin  ls,i . (2)
regional system to serve the terminal loads. Owing to the s.t. Vin = Als,i
in V
ls,i

mutual alternatives among different energy carriers and the

From the formulation, Ωi (D ls,i ) includes all possible values
redundancy of converters’ capacity, the terminal loads can be
of the input energy vector that can be converted to meet the
satisfied with a variety of combinations of multi-energy inputs.
terminal multi-energy demand of the local system without
For instance, one may consider a local system of which the violating any operating constraints of the system. Embedding
electricity demand is served simultaneously by the utility grid the flexibility of the local system in the regional system
and a local gas-fired CHP unit, while the heat load is served optimal scheduling will ensure that the scheduling results are
only by the CHP. Assume that the CHP is an extraction con- executable for the local system.
densing unit, of which the electricity-heat ratio is adjustable
within a certain range. The local system imports electricity
and natural gas from the regional system and converts them C. Coordination Scheme
into electricity and heat to serve the local demand. Since the With the explicit representation of the IDR feasible region,
terminal loads are fed by different sources, the local system the regional and local systems can be optimally coordinated
can use different combinations of electricity and natural gas through the following scheme.
from the utility system to feed a fixed amount of its terminal 1) The RSO makes an optimal schedule of system produc-
load. This feature may bring both economic and environmental tion in advance—for example, a day-ahead schedule according
benefits to the entire system. When the regional electric grid is to its load forecast of each energy node and local energy
congested or in a state of emergency, the local energy system system.
can shift from grid-supplied electricity to local CHP-generated 2) Each local multi-energy system estimates its feasible
electricity, which may help alleviate regional transmission region based on its forecasting of the terminal energy demand,
congestion or emergencies. On the other hand, when the and then submits it to the RSO at certain intervals.
regional renewable generation is in a surplus, the local system 3) According to the contingency type, real-time load and
can input more electricity from the regional system and reduce feasible region of local systems’ response, the RSO opti-
its gas input, which is helpful to accommodate renewable mally schedules utility-level systems and determines the multi-
generation and reduce emissions from fossil fuels. energy inputs of responsive local energy systems.
To embed the integrated flexibility provided by local sys- Let v rs and srs denote the vector of multi-energy flows and
tems into the regional system optimization, the capability of rs
state variables of the regional system, respectively. Let vin
the integrated flexibility provision has to be explicitly charac- ls
denote
n the resource o input of the regional system. Let Vin =
terized and submitted to the RSO. In this paper, we define the ls,1 ls,Nls
Vin , ..., Vin , where Nls denotes the number of local
feasible region of IDR as the allowable range of multi-energy
systems. Let vin and srs,p
rs,p
in respectively represent the resource
flexibility that a local system can provide to the regional
input and state variables decided by RSO in stage one, which
system without violating internal operating constraints and
can be a day-ahead dispatch result, for example. Let C rs (·)
curtailing its terminal energy demand. To illustrate the basic
and h(·) denote the cost function and the operating constraints
framework of the integrated flexibility provision, the feasible
of the regional system. Then, the optimal scheduling model
region is modelled in a compact form in this section, while
solved by the RSO is as follows,
detailed models and estimation methods will be elaborated
in Section III. Regard each local multi-energy system as an min rs
rs rs
C rs (vin ,s ) (3)
v rs ,srs ,vin
energy hub with multiple inputs and outputs. For the ith
local system, let V ls,i denote the vector of its internal energy s.t. h(v rs , srs , vin
rs ls
, Vin ) ≤ 0, (4)
flows. Let Als,i ls,i
in and Aout denote the incidence matrices of
rs
|vin rs,p
− vin | ≤ ε, (5)
the input and output ports to the internal energy flow vector, |srs − srs,p
in in | =0 (6)
respectively. Then, the input and output energy flow vectors
ls,i
of the local system can be represented as Vin = Als,i
in V
ls,i
Constraints (5) and (6) enforce corrective control constraints
ls,i ls,i ls,i
and Vout = Aout V , respectively. Given the terminal load that state variables such as generators’ decision to start up or
demand D ls,i , the operating feasible region of the local system shut down should be fixed while their output is allowed to
is denoted as Φ(D ls,i ) and can be represented in the following change within a given range ε to consider the N-1 contingency
compact form, or load fluctuating situation in the RSO’s real-time scheduling
stage. In this way, the flexible resources of local systems
 ls,i ls,i
can be considered and dispatched by the RSO, thus reducing
 
i ls,i ls,i
 g (V ) ≤ 0,
Φ (D )= V  ls,i
 A V ls,i = D ls,i , (1) contingency impact and potential line investment.
out
 This article has been accepted for publication in a future issue of this journal, but has not been edited. Content will change prior to final publication.

TONG et al.: EXPLOITING FLEXIBILITY OF INTEGRATED DEMAND RESPONSE TO ALLEVIATE POWER FLOW VIOLATION DURING LINE TRIPPING... 5

The above coordination scheme can be implemented in a follows, after which the feasible region of energy input of the
distributed fashion, i.e., the RSO does not have to collect local system can be derived through (2),
detailed information of all local systems or get control access   
i ls,i ls,i 
 (7) − (9),
to local devices. Instead, the RSO only has to obtain the Φ (D ) = V  Als,i V ls,i = D ls,i , (10)
external characteristics of local systems and determine the out
inputs needed by local systems. Compared to the integrated
optimization of the regional and local systems, the proposed B. Network Model
coordination scheme is more acceptable in practice, where The network model mainly consists of the steady-state op-
internal information and control access of local systems are eration characteristics of the electricity, gas and heat systems,
hardly open to the RSO. including their production, transmission and consumption pro-
In the proposed scheme, the first step is usually applied in cesses. The electricity network constraints are similar to the
a day-ahead way to determine the day-ahead schedule, while unit commitment problem, which are shown as follows,
the last two steps are used in real-time dispatch where the X X
system operator collects local system information and utilizes Pgt = PLnt , ∀t ∈ T (11a)
the flexibility of local systems. The above three steps can g∈U G n∈N P
also be coordinated together in a day-ahead way to perform X
a security-constrained schedule, if all kinds of contingencies (Pg,max − Pgt ) · Ugt ≥ SRtr , ∀t ∈ T (11b)
are taken into consideration in Step 3 and the dispatch results g∈U G
of Step 1 are requested to guarantee a feasible solution under
all contingency situations within the ability of the production −Fk,max ≤ Pkt ≤ Fk,max , ∀k ∈ LP , t ∈ T (11c)
and local systems to adjust. X X
L
Pkt = GSDFk,N (g) Pgt − GSDFk,n Pnt ,
g∈G n∈N P (11d)
III. M ATHEMATICAL M ODEL ∀k ∈ LP , t ∈ T
A. Local System Model
Pg,min Ugt ≤ Pgt ≤ Pg,max Ugt , ∀g ∈ U G , t ∈ T (11e)
Equations (1) and (2) give a compact form of the integrated
G
flexibility region of a local system while its specific derivation |Pgt − Pgt−1 | ≤ Rampg , ∀g ∈ U , t ∈ T (11f)
depends on the characteristics of the local system, which will ton
g,min
X
be further explained in this section. Ug,t−i ≥ ton
g,min · Zgt , ∀g ∈ U G , t ∈ T (11g)
The local system model, which consists of production i=1
components and converters of different energy systems, is tof f
g,min
illustrated by the energy conversion and local security con-
X
Ug,t−i ≤ tof f
g,min · (1 − Ygt ) , ∀g ∈ U G , t ∈ T (11h)
straints proposed in our previous work [18] as follows, i=1
ls,i ls,i ls,i Equation (11a) enforces the system power balance; equa-
Hm Am V = 0, (7)
tion (11b) shows the system reserve requirement. Equations
Qls,i ls,i ls,i
m Am V ≤ ls,i
qm , (8)
ls,i
(11c) - (11d) are the network transmission constraints. Equa-
ls,i
0 ≤V ≤V (9) tions (11e) - (11h) are the unit output and minimum start-
up/shutdown time constraints, which are enforced through 0-1
The energy conversion constraint (7) is formulated by the integer variables Ugt , Ygt and Zgt .
ls,i
energy conversion matrix Hm , which illustrates the energy The gas constraints are also based on the steady-state oper-
conversion efficiency of node m in the ith local system. Matrix ation characteristics, mainly determined by gas node pressure
Als,i
m represents the coupling matrix of ports of node m and and the flow through gas wells, pipelines and compressors,
energy flows, whose elements 1 and -1 represent that the port is which are modelled as equations (12a)-(12e). The time sub-
the sink and source of the branch of energy flow, respectively, script is omitted here, as no temporal coupling exists in the
and 0 represents that the port is not connected to the branch. constraints.
Similarly, the security constraint (8) is derived from the X X X
operating constraints of local energy converters, including vs − Ll − fmn
gs gn
capacity limits, coupled electricity-heat output constraints of s∈Um gl
l∈Um n∈Nm
(12a)
the extraction condensing CHP, etc. The coefficient matrix of
X
− Ff,j = 0, ∀m ∈ N G
node m in the ith local system Qls,i m and vector qm are
ls,i
gc
j∈Um
involved in the expression to form the constraint. The last
ls,i p
security constraint (9) enforces transfer capacity limits V fmn = sgn (πm − πn ) Cmn 2 − π2 |
|πm (12b)
n
and unidirectionality of energy flows. The specific implication
Hj
and definition of the abovementioned matrix and model are fmn = sgn (πm − πn )  α (12c)
max(πm ,πn )
involved in [20]. kj2 − kj1 min(πm ,πn )
From the above constraints, the feasible region of the energy
flows of the local system in (1) can be formulated in detail as Ff,j = cj + bj Hj + aj Hj2 , ∀j ∈ U gc (12d)
 This article has been accepted for publication in a future issue of this journal, but has not been edited. Content will change prior to final publication.
6 JOURNAL OF MODERN POWER SYSTEMS AND CLEAN ENERGY, VOL. XX, NO. XX, XXXX

πm,min ≤ πm ≤ πm,max , ∀m ∈ N G gas contract here is assumed to be a flexible contract which

vs,min ≤ vs ≤ vs,max , ∀s ∈ U gs (12e) only fixes the gas price in advance.
Hj,min ≤ Hj ≤ Hj,max , ∀j ∈ U gc
Lit = afi × Pit2 + bfi × Pit + cfi × Uit ,


The above model is a classical gas system model, which (14)
∀i ∈ U GF , t ∈ T
refers to [21]. Equation (12a) assembles the nodal balance
of gas flow. Equations (12b) - (12c) enforce the amount Equation (15) shows the relationship between the gas input
of gas flow through a pipeline or a compressor. Equation and heat output of gas-fired generators, where the coefficent
(12d) derives the loss of gas flow through a compressor. Ciboiler reflects the energy conversion efficiency from gas to
Equation (12e) shows the capability or operating range of heat. The cost here is also determined by the cost of gas
certain components, including the nodal pressure, gas supply consumption.
capability and compressors’ horsepower.
Qit = Ciboiler × Lit , ∀i ∈ U GB , t ∈ T (15)
The heat constraints are derived from the steady-state phys-
ical property between the flow mass and the heat transmission The above gas costs are determined by the price of gas
through the heat pipeline [22]. To simplify the problem, we multiplied by gas consumption, as equation (16) shows. The
assume that heat transmission is adjusted through the fluid’s set U gl here not only involves the gas load of gas boilers and
temperature instead of its flux. Additionally, the time-delay gas-fired generators, but also contains that of local systems.
property is ignored in this paper. The heat network model Wit = Lit × ρi , ∀i ∈ U gl , t ∈ T (16)
is thus described as follows, and the time subscript is also
omitted here. Equation (17) shows the relationship between the power and
heat output of CHP generators. In this way, we assume that
P 
s/b
 P 
s/b CHP works in the backpressure mode when its power output
p∈LH,+ mp Tp, out = p∈LH,− mp Tq is in direct proportion to heat output.
q
s/b s/b
q
(13a)
Tp, in = Tq , ∀p ∈ LH,− ,q ∈ NH
q Qit = Pit × CiCHP , ∀i ∈ U CHP , t ∈ T (17)
  λLp
s/b s/b −
Tp,out = Tp,in − Ta e cmp + Ta In a coordinated system, the load in different energy sec-
λL (13b) tions is divided into three categories, including fixed load,
− cmpp λL
e ≈ 1 − cmpp , ∀p ∈ LH fluctuating load and IDR load. The first type of load is given
as a forecast value, and the second type is supposed to be
Qs/1 Tps − Tpb , ∀p ∈ LHs/Hl ∩ LH


q = cmq q (13c)
available within a given range, while the last type enforces
s s s H,s
Tmin ≤ Tp,out/in ≤ Tmax , ∀p ∈ L the constraint that the electricity, heat and gas consumption
b b b (13d)
Tmin ≤ Tp,out/in ≤ Tmax , ∀p ∈ LH,b connected to the same SEH vary in a given feasible region
determined by its physical characteristics, as stated in Section
Equation (13a) describes the procedure of heat mixing at
III-A.
node q, representing the heat mix balance constraint and the
In day-ahead scheduling, for example, given the load fore-
nodal outflow temperature constraint. Equation (13b) models
cast information, the complete model to minimize the opera-
the heat loss during the transmission procedure through a
tion cost of the regional system while satisfying the system
pipeline, reflected in the difference between the temperature
load can be written as follows,
of the fluid outflow and inflow of the pipeline, which can be  
P Q G
further linearized to simplify the calculation. Equation (13c) M in Obj = f vin , vin , vin
shows the relationship between the heat supply or load and (18)
s.t. (2), (10) − (17)
the fluid flux combined with its temperature. Equation (13d)
enforces the temperature limit of the network. The objective function in (18) is related with the multi-
P Q G
The proposed network model adopts a DCOPF-based power energy input of the regional system vin , vin , vin , which can
system model and neglects the time-delay property of the heat be expanded to the sum of the cost of each energy produc-
system. Future studies may attempt present a more elaborate tion process, as shown by (19). Here, fiCHP (·) and fiCF (·)
model, such as a distribution network model. represent the quadratic cost function of coal-fired CHPs and
thermal generators, as the costs of gas-fired CHPs and thermal
C. Coordinated Optimization generators are included in the gas costs. SUit and SDit
represent the start-up and shut-down costs of generators,
Different energy systems are connected through the energy respectively.
production and consumption processes. The consumption pro- 
cess includes the integrated flexibility the local system can X X X
provide, whose model has been introduced in section III-A. f= Wit + fiCHP (Pit , Qit ) Uit
 gl
This section will mainly introduce the model used in other t i∈U i∈U CHP
) (19)
energy conversion processes. X X
CF
Equation (14) shows the relationship between the gas input + fi (Pit ) Uit + [SUit + SDit ]
and power output of gas-fired generators, where gas consump- i∈U CF i∈U G

tion is modelled as a quadratic function of the power output, By introducing a large penalty factor of the slack variable of
whose coefficients are represented by afi , bfi and cfi . The line flow, the impact of the N-1 contingency on the violation
 This article has been accepted for publication in a future issue of this journal, but has not been edited. Content will change prior to final publication.

TONG et al.: EXPLOITING FLEXIBILITY OF INTEGRATED DEMAND RESPONSE TO ALLEVIATE POWER FLOW VIOLATION DURING LINE TRIPPING... 7

of transmission capacity constraints can be studied. The slack

variables, representing the maximum power flow violation of
ϰ ϳ '^ϭ
power transmission lines, will remain zero when no power flow ‫ܮܩ‬ଵ  ϭ
‫଻ܮܩ‬
‫଺ܮܩ‬
‫ܮܩ‬ହ

violation occurs. However, if violation is inevitable when the ‫ܮܩ‬ଶ

Ϯ ଵ 

‫଼ܮܩ‬
adjustment ability of IDR and generators is insufficient, the ‫ܮܩ‬ଷ ϯ ϱ ϲ '^Ϯ

slack variable will be exactly the maximum violation value of ‫ܮܩ‬ସ 

the line capacity. In this way, the objective function can be

rewritten as (20),

X X X X ‫ܪ‬ଶ

f= Wit + fiCHP (Pit , Qit ) Uit + ,W '

 gl ‫ܵܪ‬ϭ ‫ʹܵܪ‬
t i∈U i∈U CHP i∈U CF
) ‫ܪ‬ଵ
X X 'ĂƐĨŽƌŐĂƐͲĨŝƌĞĚŐĞŶĞƌĂƚŽƌƐ
fiCF (Pit ) Uit + [SUit + SDit ] + σ slk 'ĂƐĨŽƌŚĞĂƚďŽŝůĞƌƐ
,ĞĂƚƉƌŽĚƵĐĞĚďLJ,W
i∈U G k∈LP
(20) Fig. 2. System topology of the proposed multi-energy system.
At the same time, the line flow constraints (11c) should be
rewritten as follows, 3 pipelines connecting them. The heat sources are connected
−Fk,max − slk ≤ Pkt ≤ Fk,max + slk , slk ≥ 0 (21) with CHP generators and gas boilers whose heating capacities
are 130 MW and 100 MW, respectively, while the heat loads
In this paper, only power system transmission line tripping are assumed to be the sum of the fluctuating load within a
contingencies and their impact are discussed. Contingency and given range and SEH load. The three SEHs are connected
its impact in other systems may be further studied in the future. to gas node 3 and power buses B4, B5, and B8 and heat
The entire model is a mixed-integer nonlinear programming node HL1 in the three heat networks. Each local SEH contains
problem, while after applying piecewise linearization in [23] an electric transformer, a gas-fired CHP, a gas boiler and an
to the gas flow constraints, it can be transformed into a linear electric heat pump, whose rated capacities are 15 MW, 15 MW,
model, which can be solved using mainstream optimization 7.5 MW and 3 MW, respectively. The system-wide information
solvers. of the proposed case and system load curve are shown in Table
I and Fig. 3. The standardized load of each energy carrier
IV. C ASE S TUDY equals the hourly load divided by its daily maximum load.
In this section, a multi-energy system containing electricity,
gas and heat networks, in which different energy sources and TABLE I
local integrated energy conversion models are embedded, is S YSTEM - WIDE INFORMATION OF THE PROPOSED CASE
proposed. The benefits of applying local system flexibility to
Parameters Total capacities Maximum load Production cost
alleviate the impacts caused by N-1 power system transmission
Electricity system 4705 MW 2850 MW 41∼128$/MWh
line tripping contingencies are analyzed based on the case Gas system 11.3 Mcf/h 6.64 Mcf/h 6.23$/Mbtu
system. Heat system 230 MW 155 MW /
1 The heat production cost is assumed as the cost of the consumed electricity
A. System Description and gas.
2 1 kilo-cubic feet of natural gas is assumed to generate 1 MBtu of energy.
The proposed system is based on a modified 24-node IEEE
The case study is formed on a daily basis, while the time
RTS96 power system, together with a 7-node gas system and
interval is set to one hour. The full optimization problem,
three independent 4-node heat systems. The components and
as expression (18) states, is a mixed-integer nonlinear pro-
topology of the multi-energy system are shown in Fig. 2,
gramming problem that can be reformulated into a mixed-
In the proposed power system, three of the generators
integer linear programming (MILP) problem by adopting
are replaced by gas-fired CHP generators, and renewable
piecewise linearization methods, as shown in the Appendix.
resources, including wind and solar energy, are added in
The simplified MILP problem is modelled by GAMS 24.3
the top half of the system. The total installed capacities of
using Cplex 12.6 as the solver on a Thinkpad T490 laptop.
thermal generators, gas-fired CHPs, and renewable generators
are 3153 MW, 252 MW and 1300 MW, respectively. The
1
load of the power system includes fixed residential load and 0.9
electricity
gas
heat
load connected to the SEH. The 7-node gas system, whose 0.8

0.7
model data refer to [21], contains 2 gas wells, 5 gas pipelines
Standardized load

0.6

and 1 compressor. The total gas production capacity is 11.3 0.5

0.4
Mcf/h. The load of the gas system includes the fixed residential 0.3

load (GL7), SEH load (GL8) and consumption of gas boilers 0.2

and gas-fired generators (GL1~GL6), which are assumed as 0.1

0
flexible contracts, depending on their power/heat output. The 0 5 10
Time (h)
15 20 25

heat system consists of 2 heat sources and 2 heat loads, with Fig. 3. Standardized system load curve.
 This article has been accepted for publication in a future issue of this journal, but has not been edited. Content will change prior to final publication.
8 JOURNAL OF MODERN POWER SYSTEMS AND CLEAN ENERGY, VOL. XX, NO. XX, XXXX

B. Illustration of the feasible region of IDR 100 100

(a) (b) Power flow in S1
The results of the IDR feasible region of a local SEH Power flow in S2
Power flow in S3
at certain time intervals are shown in Fig. 4. The feasible 50 50 Line capacity

region reflects the local system’s ability to change its load

Power flow (MW)

Power flow (MW)

components as the inputs of SEH can be an arbitrary vector 0 0

within the region to reduce the system’s real-time cost or

respond to the system’s order to increase its security margin. -50 -50

For example, the operation point of the IDR can move from
-100 -100
an interior point to its border to reduce certain loads when
certain equipment is in outage in any energy system.
-150 -150
5 10 15 20 5 10 15 20
Time (h) Time (h)
Fig. 5. Power flow in S2 and S3 when line 27 is tripping. (a) Flow of line
6. (b) Flow of line 10.

300 750 280

(a) (b) (c) S2
260 S3
700
250
240
650

Electricity load (MW)

220
200

Heat load(MW)
Gas load (MW)
600
200

150 550 180

Fig. 4. IDR feasible region of a local SEH. (a) At time interval 9. (b) At
time interval 19.
C. Effectiveness of the Coordination Framework
In this section, 3 scenarios are considered: S1, normal
operation without IDR; S2, operation during N-1 contingency
without IDR; and S3, operation during N-1 contingency with
IDR. By comparing the slack variable of line capacity and
introducing a large penalty term in the objective function, the
effectiveness of IDR in alleviating flow violation caused by
transmission line tripping is illustrated. The system optimiza-
tion procedure during line tripping contingencies is similar to
the methods used in corrective control, in which the start-up
and shut-down decisions of the units are fixed to the result of
the day-ahead unit commitment in scenario S1, but the power
outputs are allowed to change within a given range.
Taking a single line tripping condition as an example, as-
sume that line 27 from node 15 to node 24 is in outage during
a whole day. Then, the day-ahead optimization of the proposed
multi-energy system operation in three scenarios is performed.
The result indicates that the line tripping contingency will
make the system infeasible during certain time intervals in
scenarios S2 & S3 due to the lack of power transmission
capacity, and the power flow of certain transmission lines in
S2 and S3 is demonstrated in Fig. 5.
From Fig. 5, the power flow of lines 6 & 10 in S2 will
exceed the lower bound in hours 9, 18 and 19, while the
power flow constraints during these hours are all redundant in
S1, which implies that line tripping increases the load rate of
lines 6 & 10 and makes the line exceed its maximum capacity
and thus the whole system infeasible in S2. However, the
violation of power transmission constraints is greatly reduced
in scenario S3 when IDR is considered, as only the power flow
of line 6 in hour 19 exceeds its limit by 12.3 MW compared
with the 31.1 MW in S2 due to the load adjustment of the
power system.
160
500
100
140
450
120
50
400
100
0 350 80
5 10 15 20 5 10 15 20 5 10 15 20
Time (h) Time (h) Time (h)
Fig. 6. Total load of local systems. (a) Electricity load. (b) Gas load. (c) Heat
load.
The total energy load of the local systems participating
in IDR is shown in Fig. 6, from which the shedding of
electricity load can be seen during these time intervals in
S3 compared with those in S1 and S2. More specifically, the
local system uses more gas and heat instead of electricity to
satisfy the terminal energy requirement, thus providing more
feasibility and a greater security margin for power system
during power transmission contingencies. If the LTE/STE
of the transmission line is considered in the coordination
operation procedure, the adoption of IDR can also reduce the
requirement of LTE/STE, thus making the whole system safer.
To be more specific, the system-wide result of each line
tripping condition, including the feasibility with and without
IDR and the system’s total cost, is shown in Table II. Among
all 37 line tripping conditions (with one ignored because it
will cause a ‘power island’), the adoption of IDR can prevent
the infeasibility that line tripping may bring about in 21
conditions. In 8 conditions, the entire system is feasible in
both S2 and S3, where IDR can further reduce the system’s
total cost by making terminal loads more rational according to
the energy production or transmission situation. For example,
the IDR can help to consume renewable energy when there
is excessive wind output. In other conditions, such as the
abovementioned line 27 tripping situation, taking IDR into
consideration can reduce transmission line capacity violations.
After a single line tripping case, all possible line tripping
conditions are studied together, among which the maximum
and minimum values of the power flow of each line in each
time interval are then derived from the optimization results
to show the power transmission feasibility change caused by
IDRs. Then, the decrease in maximum slack required by each
 This article has been accepted for publication in a future issue of this journal, but has not been edited. Content will change prior to final publication.

TONG et al.: EXPLOITING FLEXIBILITY OF INTEGRATED DEMAND RESPONSE TO ALLEVIATE POWER FLOW VIOLATION DURING LINE TRIPPING... 9

TABLE II
C ASE RESULT OF ALL LINE TRIPPING CONDITONS

Line Outage 5,7,9,16∼18,23,27 4 6 24 28 30 31 32 33 else

Feasibility without/with IDR 0/0 1/1 1/1 1/1 1/1 1/1 1/1 1/1 1/1 0/1
Total cost without IDR(106 $) 2.705 2.706 2.708 2.705 2.705 2.714 2.705 2.705
Total cost with IDR(106 $) 2.667 2.667 2.669 2.666 2.666 2.677 2.667 2.667
1 0 represents system infeasiblity while 1 represents feasibility.
2 Line 11 tripping situation is ignored.

line through S2 to S3 can also provide a way to quantify the V. C ONCLUSION

IDR’s benefits in terms of system security. The comparison of To quantify the flexibility of local energy systems and its
the extremum of the power flow of lines 12 and 13 in S2 and effect on system operation during contingencies, this paper
S3 is given in Fig. 7. first proposes a feasible region model of IDR to depict the
energy consumption and conversion process of a local system
100 100
and derive the flexibility of its energy input. The feasible
(a) (b) minimum/maximum
power flow in S2
region derived can be embedded into the system scheduling
50 50 minimum/maximum
power flow in S3 process without detailed information of local system equip-
line capacity
ment to ensure privacy. A centralized coordination scheme
Power flow (MW)

Power flow (MW)

0 0
and optimization model of the entire system is then proposed,
-50 -50
based on which the impact of multi-energy flexibility on
-100 -100
the power system security margin, N-1 power system line
tripping feasibility in particular, is further studied. Through
-150 -150
the multi-energy system case, it can be seen that the proposed
-200 -200 scheme involving IDR of local systems can enhance the
5 10 15 20 5 10 15 20
Time (h) Time (h) power system reliability towards N-1 transmission line tripping
Fig. 7. Extreme power flow in all conditions. (a) Line 12. (b) Line 13. contingencies. The proposed method can be further extended
and applied to other energy systems in multi-energy systems to
In Fig. 7, the blue dotted line represents the minimum and assess their stability in the presence of multiple contingencies.
maximum power flow in all line tripping conditions in S2
while the red full line represents the extreme power flow in S3. A PPENDIX A
From time intervals 7 to 23, the extreme power flows of lines
12 and 13 both exceed the lower bound in S2. However, by For nonlinear expression h(x) and x1 < x2 < ... < xn
adopting IDR instead of the fixed load, the slack variable these in its domain, the linearized process can be stated as follows,
lines require can be reduced from about 71.4 MW to zero, and where δi is a continuous variable and yi is a binary variable
thus the power flow violation is eliminated when reflected in [23].
the figure, which means that the flexibility provided by IDR n−1
X
contributes to the secure operation of these lines in any N-1 h(x) ≈ h (x1 ) + (h (xi+1 ) − h (xi )) δi (A1)
contingencies. i=1

n−1
X
TABLE III x = x1 + (xi+1 − xi ) δi (A2)
C ASE RESULT OF POTENTIAL TRANSMISSION LINE VIOLATION i=1

Number of Total violated time Maximum flow δi+1 ≤ yi , yi ≤ δi , i = 1, 2, ..., n − 1 (A3)

Index
line violations intervals of lines violation (MW)
S2 15 141 108.29
0 ≤ δi ≤ 1, i = 1, 2, ..., n (A4)
S3 8 44 45.09
The nonlinear constraints in this paper mainly consist of the
Average flow Maximum flow Average flow
violation (MW) violation rate(%) violation rate(%)
following three constraints: the gas consumption constraints
of gas-fired generators, gas transmission line constraints and
52.29 72.64 31.23
24.47 38.33 16.41 gas compressor constraints. In the gas consumption equation
1
(14), the quadratic component can be linearized through the
The average value is based on the value of violated lines in each
condition. above expression (A1) – (A4) by letting h(x) denote the square
function and x denote the power output.
Table III depicts the overall power flow violation condition In the gas flow constraint (12b), we assume that for each
in N-1 situations. The adoption of IDR reduces the number of pipeline, the flow direction is previously determined to elimi-
transmission lines suffering potential violations from 15 to 8 nate the absolute value. It can also be addressed by introducing
and the total violated time intervals of lines from 141 to 44. a binary variable to compare the pressure of the two ports. If
The maximum and average power flow violations also greatly we assume that πm > πn , the gas flow constraints can be
decline with IDRs. linearized as follows by using the square of pressure as the
 This article has been accepted for publication in a future issue of this journal, but has not been edited. Content will change prior to final publication.

10 JOURNAL OF MODERN POWER SYSTEMS AND CLEAN ENERGY, VOL. XX, NO. XX, XXXX

independent variable and letting h(·) represent the square root [9] J. Wang, H. Zhong, Z. Ma et al., “Review and prospect of integrated
calculation and x1 equal 0. demand response in the multi-energy system,” Applied Energy, vol. 202,
pp. 772–782, 2017.
n−1 [10] Z. Chen, Y. Sun, X. Ai et al., “Integrated demand response character-
X √ √
istics of industrial park: A review,” Journal of Modern Power Systems
fmn ≈ Cmn [ xi+1 − xi δ i ] (A5)
and Clean Energy, vol. 8, no. 1, pp. 15–26, 2020.
i=1
[11] S. Zheng, Y. Sun, B. Li et al., “Incentive-based integrated demand
n−1
X response for multiple energy carriers considering behavioral coupling
2
πm − πn2 = (xi+1 − xi ) δi (A6) effect of consumers,” IEEE Transactions on Smart Grid, vol. 11, no. 4,
pp. 3231–3245, 2020.
i=1 [12] S. Bahrami and A. Sheikhi, “From demand response in smart grid toward
δi+1 ≤ yi , yi ≤ δi , i = 1, 2, ..., n − 1 (A7) integrated demand response in smart energy hub,” IEEE Transactions on
Smart Grid, vol. 7, no. 2, pp. 650–658, 2016.
0 ≤ δi ≤ 1, i = 1, 2, ..., n (A8) [13] C. Shao, Y. Ding, P. Siano et al., “A framework for incorporating demand
response of smart buildings into the integrated heat and electricity energy
In the compressor gas flow constraint (12c), we use a step system,” IEEE Transactions on Industrial Electronics, vol. 66, no. 2, pp.
1465–1475, 2019.
function to estimate the compression ratio πm /πn . Addition- [14] F. Dababneh and L. Li, “Integrated electricity and natural gas demand
ally, we apply the same assumption in the gas flow equation response for manufacturers in the smart grid,” IEEE Transactions on
and use the square of pressure as the independent variable. Smart Grid, vol. 10, no. 4, pp. 4164–4174, 2019.
[15] P. Liu, T. Ding, Z. Zou et al., “Integrated demand response for a load
The estimation is shown as follows, where Ri represents a serving entity in multi-energy market considering network constraints,”
series of estimated values in the allowed range, yi represents Applied Energy, vol. 250, pp. 512–529, 2019.
a binary variable to choose a Ri as the square of the ratio. [16] D. Chen, C. Wan, Y. Song et al., “N-1 security-constrained coordinated
scheduling of integrated electricity and natural gas system considering
n
X n
X gas dynamics and wind power uncertainty,” IET Renewable Power
2
πm /πn2 = Ri yi , yi = 1 (A9) Generation, vol. 15, no. 7, pp. 1408–1421, 2021.
[17] H. Chen, J. Shao, T. Jang et al., “Static N-1 security analysis for inte-
i=1 i=1
grated energy system based on decoupled multi-energy flow calculation
1 method,” Automation of Electric Power Systems, vol. 43, no. 17, pp.
Let parameter Coni = α/2 . Then the compressor flow
kj2 −kj1Ri 20–35+, 2019.
can be expressed as follows, [18] Z. Tan, H. Zhong, Q. Xia et al., “Exploiting integrated flexibility from a
n local smart energy hub,” in 2020 IEEE PES General Meeting, Montreal,
X 2020, pp. 1–5.
fmn = Coni (yi Hj ) (A10) [19] M. Geidl and G. Andersson, “Optimal power flow of multiple energy
i=1 carriers,” IEEE Transactions on Power Systems, vol. 22, no. 1, pp. 145–
155, 2007.
The above expression involves the product of a binary variable [20] Y. Wang, N. Zhang, C. Kang et al., “Standardized matrix modeling of
and a continuous variable and still requires some skill to be multiple energy systems,” IEEE Transactions on Smart Grid, vol. 10,
transformed into a linear expression. Let G = Y X, which no. 1, pp. 257–270, 2019.
[21] C. Liu, M. Shahidehpour, Y. Fu et al., “Security-constrained unit com-
represents the value of the product of a binary variable Y and mitment with natural gas transmission constraints,” IEEE Transactions
a continuous variable X. By introducing a large parameter M , on Power Systems, vol. 24, no. 3, pp. 1523–1536, 2009.
it can be expressed in a linear form as follows, [22] M. Pirouti, “Modelling and analysis of a district heating network,” Ph.D.
dissertation, Cardiff University, 2013.
−M Y ≤ G ≤ M Y (A11) [23] C. M. Correa-Posada and P. Sánchez-Martín, “Gas network optimization:
A comparison of piecewise linear models,” 2014. [Online]. Available:
−M (1 − Y ) ≤ X − G ≤ M (1 − Y ) (A12) http://www.optimization-online.org/DB_FILE/2014/10/4580.pdf

R EFERENCES Tong Cheng received the B.S. degree in electrical engineering

[1] B. Stott, O. Alsac, and A. J. Monticelli, “Security analysis and opti-
mization,” Proceedings of the IEEE, vol. 75, no. 12, pp. 1623–1644,
1987.
[2] K. W. Hedman, M. C. Ferris, R. P. O’Neill et al., “Co-optimization
of generation unit commitment and transmission switching with N-1
reliability,” IEEE Transactions on Power Systems, vol. 25, no. 2, pp.
1052–1063, 2010.
[3] D. A. Tejada-Arango, P. Sánchez-Martın, and A. Ramos, “Security
constrained unit commitment using line outage distribution factors,”
IEEE Transactions on Power Systems, vol. 33, no. 1, pp. 329–337, 2018.
[4] Y. Wang, Q. Xia, M. Zhu et al., “Optimal corrective security constrained
unit commitment model and algorithm,” Automation of Electric Power
Systems, vol. 35, no. 9, pp. 19–24, 2011.
[5] S. Maslennikov and E. Litvinov, “Adaptive emergency transmission rates
in power system and market operation,” IEEE Transactions on Power
Systems, vol. 24, no. 2, pp. 923–929, 2009.
[6] Y. Fu, M. Shahidehpour, and Z. Li, “AC contingency dispatch based
on security-constrained unit commitment,” IEEE Transactions on Power
Systems, vol. 21, no. 2, pp. 897–908, 2006.
[7] C.-L. Su and D. Kirschen, “Quantifying the effect of demand response
on electricity markets,” IEEE Transactions on Power Systems, vol. 24,
no. 3, pp. 1199–1207, 2009.
[8] Y. Wang, I. R. Pordanjani, and W. Xu, “An event-driven demand
response scheme for power system security enhancement,” IEEE Trans-
actions on Smart Grid, vol. 2, no. 1, pp. 23–29, 2011.
from Tsinghua University, Beijing, China, in 2020. He is
currently working toward the Ph.D. degree in electrical
engineering at Tsinghua University. His research interests
include power system operation and multi-energy systems.
Zhenfei Tan received the B.S. degree in electrical engineering
from Tsinghua University, Beijing, China, in 2017. He is
currently pursuing the Ph.D. degree in electrical engineering
with Tsinghua University. He serves as the Vice-Chair of
the IEEE PES Tsinghua Student Branch Chapter since 2020.
He was a recipient of the “Best-of-the-Best” Conference
Paper Award in the IEEE PES General Meeting in 2020. His
research lies in the distributed and coordinated optimization
of power systems.
Haiwang Zhong received the B.S. and Ph.D. degrees in
electrical engineering from Tsinghua University, where he
is currently an Associate Professor with the Department of
Electrical Engineering. He was a recipient of the ProSPER.
 This article has been accepted for publication in a future issue of this journal, but has not been edited. Content will change prior to final publication.
TONG et al.: EXPLOITING FLEXIBILITY OF INTEGRATED DEMAND RESPONSE TO ALLEVIATE POWER FLOW VIOLATION DURING LINE TRIPPING... 11

Net Young Scientist Award in 2014, the Outstanding Ph.D.

Graduate and Outstanding Post-Doctor of Tsinghua University.
He serves as the Editor for the International Transactions
on Electrical Energy Systems, an Associate Editor for the
CSEE Journal of Power and Energy Systems. He currently
serves as the Chair of the Demand Response Working Group
under the IEEE PES Power System Operations, Planning and
Economics Committee, the Secretary of the IEEE PES SBLC
Architecture Subcommittee, and the Secretary of CIGRE
D2.53 Working Group on Technology and Applications of
Internet of Things in Power Systems. His research interests
include smart grids, power system operations, electricity
markets, and demand response.