Review on Energy Resilience

Energy Policy 146 (2020) 111795

Available online 2 August 2020 0301-4215/© 2020 Elsevier Ltd. All rights reserved.

Resilience of China’s oil import system under external shocks: A system dynamics simulation analysis

Sai Chen a, b, Ming Zhang a, b, *, Yueting Ding a, b, Rui Nie a

a School of Economics and Management, China University of Mining and Technology, Xuzhou, 221116, China b Center for Environmental Management and Economics Policy Research, China University of Mining and Technology, Xuzhou, 221116, China

A R T I C L E I N F O

Keywords: External shocks Resilience System dynamics (SD) Oil import system

A B S T R A C T

Oil is an important energy resource that guarantees the operation of a country’s production and life. To reduce the losses caused by the interruption of oil imports, it is urgent to build a system with resilience. By referring to the idea of resilience evolution curve and analyzing the feedback relationship among the four sub-modules of China’s oil import system, this paper established a system dynamics (SD) simulation model to study resilience of China’s oil import system under external shocks. Then, according to the different parameter groups, we simu- lated the performance changes of the oil import system in different scenarios, calculated the values of system resilience in different situations, and analyzed the possible critical points that system can still maintain normal operation. i) Diversified measures can enhance system resilience and reduce losses. ii) When the external shocks are strong, the extraordinary production of coal-to-oil and public participation also play an important role in mitigating risks. iii) The increase of strategic crude oil reserves, conversion coefficient of crude oil, the com- pany’s ability to guarantee oil security and the ratio of energy substitution enhance system resilience effectively. iv) The threshold for the system to maintain stability is found under a certain scenario.

1. Introduction

Since China became the net oil importer for the first time in 1993, the dependence of oil imports has been increasing year by year (Wu et al., 2013). In 2018, China’s net oil imports reached 461.9 million tons, and 70% of its oil consumption was obtained from foreign imports, an in- crease of 10.1% over 2017 (Zhang, 2019). According to the current situation of oil imports, the source countries of imports are mainly concentrated in the turbulent Middle East and Africa, which means that China has to import oil over an average distance of 21,000 km, through many rugged islands, canals (Strait of Malacca, Strait of Hormuz, Persian Gulf, Suez Canal, etc) and pirate areas (Bay of Bengal, Somalia, West Africa, Caribbean, etc) (Wang et al., 2018). Therefore, there are always many risks in the process of China’s oil import, such as: extreme weather, tsunami, earthquake, typhoon and other natural disasters, financial crisis, local armed conflicts, etc.(Beccue et al., 2018). In this paper, the above-mentioned risks are defined as external shocks in the process of oil import.

The high degree of external dependence and inevitable external shocks has laid many hidden dangers for the security of oil import, which may lead to serious consequences to the whole economy and

society. Since 1950, there have been 31 statistically recorded oil supply shortages worldwide, with supply disruptions ranging from 0.5 to 44.7 months (Guo, 2017). For instance, the hurricane Katrina in the US Gulf of Mexico in 2005 caused a massive shortage of oil production and refining in the United States; The Libyan War in 2011 interrupted oil supply by up to 1.6 million barrels per day. In 2018, the United States imposed sanctions on Iran and Russia, including an embargo on Iranian crude oil and a crackdown on Russia’s energy industry, which affected stable oil supplies. According to import data from the General Admin- istration of Customs, China’s oil imports from Iran fell by 49.5 percent from 292.7 million tons in 2018 to 147.8 million tons in 2019.

Energy security (risk) caused by external shocks is a mainstream topic in the energy field. Actually, a number of studies have discussed it, for example: In terms of the concept of energy security, Yao (2018) studied how to conceptualize energy security in four resource-poor advanced island countries (Singapore, South Korea, Japan and Taiwan). In terms of energy security quantification, Song et al. (2019) introduced a new comprehensive index, China energy security index (CESI), to evaluate the changes of China’s energy security over the years. Four aspects of the availability of energy resources, the applicability of technology, the acceptability by society, and the affordability of energy

* Corresponding author. School of Economics and Management, China University of Mining and Technology, Xuzhou, 221116, China. E-mail address: [email protected] (M. Zhang).

Contents lists available at ScienceDirect

Energy Policy

journal homepage: http://www.elsevier.com/locate/enpol

https://doi.org/10.1016/j.enpol.2020.111795 Received 29 May 2020; Received in revised form 19 July 2020; Accepted 26 July 2020

Energy Policy 146 (2020) 111795

2

resources were selected by Yao and Chang (2014) to establish a 4-as framework, which comprehensively quantified energy security. In terms of the influencing factors of energy security, Yao (2015) analyzed the relationship between energy security and energy policy from the perspective of macroeconomic reform. Similarly, as an important energy source, oil risk has also been studied by many scholars. For example, Sun et al. (2017) evaluated the systematic risks of China’s oil supply chain from four aspects: availability, accessibility, acceptability and afford- ability of crude oil imports. Yang et al. (2014) used the diversification index method, taking into account the country risk of the source country of import and the potential export capacity to assess the external oil supply risks of China, Japan, the United States and the European Union. Similarly, Sun et al. (2014) quantified the risk of China’s oil import by taking into account both the country risk of oil exports countries and the risks of oil transportation routes. However, above literatures mainly focused on static index, while energy (oil) import security is a dynamic process.

In recent years, against the background of frequent “black swan” and “grey rhinoceros” incidents, the thought of system resilience has become a research hotspot due to its characteristics of fully estimating risks and sticking to the bottom line. In risk management, system resilience is defined as the ability of the system to cope with risks, and focus on how to quickly restore the system to a normal operating state (Afgan and Veziroglu, 2012; Matzenberger et al., 2015). Viewed as a “shield,” a “shock absorber,” or some sort of buffer, a resilient system minimizes the vulnerability of the system, takes advantage of favorable opportunities, and ensures that the results are benign or have only a small negative impact (O Brien and Hope, 2010; Zhang and Liu, 2015). Therefore, system resilience is crucial for the system to recover quickly and main- tain basic operation in the face of external shocks. Just as the United Nations International Strategy for Disaster Reduction (UNISDR) pointed out, resilience is a valuable quality shared by human society and the natural world, and a goal that should be pursued (ISDRUN, 2004). However, as a new entry point for risk management, there are few studies on system resilience, which are scattered in other research fields. In the limited literature (Ouyang and Duenas-Osorio, 2014; Ouyang and Wang, 2015; Ouyang, 2017), established the resilience model of power grid and other infrastructure under earthquake, hurricane and other disaster scenarios by simulating the disaster situation, analyzing the vulnerability of system components, and calculating the change of the system. Amirioun et al. (2017) and Amirioun et al. (2019) described and quantified the degradation and recovery process of the micro power grid in response to the storm through the vulnerability curve. The above research results on resilience have important reference significance for applying the theory of resilience to the oil import system and con- structing the resilience model of China’s oil import system under external shocks.

Through combing the relevant resilience literature, it is found that there are fewer empirical studies than the results obtained by theoretical research. In more literatures, a resilience research framework has been established, but the resilience value has not been measured. Practically, the measurement of system resilience is not only an important part of the study of resilience, but also an important way to improve system resil- ience. The literature about resilience measurement can be roughly divided into two categories: one is the resilience measurement based on the index system. One of the representative studies is Linkov et al. (2013), According to the definition of resilience characteristics by the National Academy of Sciences, namely preparation, absorption, recov- ery and adaptation, a resilience measurement matrix of 4 � 4 is con- structed by using the four fields of physics, information, cognition and society described in the theory of Network Centric Operations (NCO). Furthermore, Roege et al. (2014) provide a more specific supplement to the evaluation matrix proposed by Linkov for energy systems. Erker et al. (2017a, b) also used the matrix method to construct a resilience matrix combining the “feature dimension” and the “domain dimension”; the difference is that in terms of index selection, Erker not only selected

fact evaluation indicators but also value evaluation indicators; However, it can be seen that, there is no indicator system that can be applied to all events (Roege et al., 2014). These papers are basically using static in- dicators to illustrate the resilience. While indicators could convey in- formation, they fail to present the process.

The other category is the resilience measurement based on the resilience evolution curve. It describes the dynamic process of perfor- mance degradation and recovery under external shocks (Kulig and Hanson, 1996; Zhou et al., 2010). In many literatures, the measurement of resilience is based on the resilience evolution curve (Henry and Ramirez-Marquez, 2012; Cimellaro et al., 2014). Omer et al. (2013)were pioneers in calculating resilience by using the area enclosed by the resilience evolution curve and the time axis. Similar studies also included Bruneau et al. (2003), who constructed the community resil- ience model under earthquake disasters by using the resilience evolution curve. Ouyang et al. (2012)also calculated resilience according to the resilience evolution curve. The difference is that the ratio of the disturbed area to the area under normal conditions is taken as the measurement standard of resilience. In all, compared with the static measurement of resilience through the indicator system, the method of measuring resilience through the evolution curve is more dynamic and systematic.

According to the above analysis, the idea of resilience evolution curve is used to construct the resilience model of China’s oil import system under external shocks. Due to the many variables involved in the system model, and there are many feedbacks and time delays in the system, it is difficult to quantify those accurately using traditional methods. However, System dynamics (SD), as a computer-aided system simulation method, has unique advantages in quantitative analysis of nonlinear and multi-feedback complex time-varying systems (Garbolino et al., 2016; Liu and Zeng, 2017), Therefore, SD is selected to depict the resilience evolution curve in this work. In recent years, SD method has been widely used in the field of energy emergency management. For example, Wang and Zhang (2019) established an evaluation model of potential mobilization of emergency supplies by using SD, and con- ducted a quantitative evaluation of potential mobilization of refined oil against the background of China’s oil industry chain. Wang and Qu (2019) simulated the resilience of the imported crude oil supply chain when the disturbance occurred by using the SD on the basis of analyzing the interference factors of the imported crude oil supply chain. Gong and Zhang (2017) built a SD model for the emergency transportation of refined oil, and took the emergency transportation of refined oil of CNPC after the WenChuan earthquake in 2008 as an example to carry out simulation verification.

Based on the above background, to clearly understand the degra- dation and recovery (dynamic process) of the China oil import system under external shocks, how resilient is China’s oil import system under external shocks? How to restore the system to a better state after the shock? What factors determine the speed that the system recovers from damage? What is the bottom line of a system? All above issues will be studied in this paper. Specifically, based on the resilience evolution curve, the resilience measurement model was established firstly; Sec- ondly, according to the SD modeling steps, the system causality diagram and the system flow diagram were constructed, and the relationship between variables was also determined. Finally, the system perfor- mances of different scenarios were simulated according to different parameters, and the system resilience value under different conditions was calculated through the evolution curve. Compared with the existing research, there are two innovations in this paper: One is clarify the conversion relationship between crude oil and refined oil, avoid confusion in the modeling process, which helps to make the system simulation results more accurate and practical. The other is the resil- ience value under different scenarios is calculated through analyzing the feedback relationship within each sub-module that constitutes the sys- tem. Furthermore, the threshold at which the system maintains opera- tion is found, these innovative works not only helps enriching the theory

S. Chen et al.

Energy Policy 146 (2020) 111795

3

of resilience but also provide new ideas and methods for the imple- mentation of energy policies.

2. System resilience measurement model based on evolution curve

Based on the above review and analysis of the domestic and foreign literatures, the following content in this section focuses on the resilience evolution curve. Judging from the dynamic evolution of the system under external shocks, the resilience measurement model is mainly composed of two stages: degradation and recovery. Specifically, the resilience evolution curve is shown in Fig. 1, during degradation, the damage degrees of the system are different, and the θ reflects three possible situations in this stage, namely U1, U2 and U3, which corre- spond to rapid decline, gradual decline and slow decline of the system performance respectively. During recovery, the recovery speed and re- sults of the system are different, and the γ reflects the three conditions: V1, V2 and V3, that is, the new steady-state may decline, stabilize and evolve compared with the initial state. Actually, although the θ and γ represent different paths of system performance, the algorithms to measure resilience are the same. As is shown in Fig. 1, where PðtÞ is a system performance function, that is, a resilience curve function. For different systems, the proxy variables of the performance function often vary according to the research object. In fact, in some studies, Multi- index synthesis functions are often used to determine the system per- formance index of resilience (Kim et al., 2015; Tan et al., 2019), which includes the relatively typical user satisfaction rate (Xu et al., 2014; Mari et al., 2015). Therefore, it is reasonable that the Demand Satisfaction Rate (DSR) of China’s oil consumption is selected as the performance function in this work. Pt0 represents the system performance value for the initial state of the system, Pt1 represents the lowest value of system performance under external shocks, t0 represents the time when an external shock occurs, t1 represents the time at which system perfor- mance hit bottom, t2 represents the time at which system recovery is completed. To sum up, after the occurrence of external shocks, the loss in the performance degradation stage of the system is:

RU ¼

R t1 t0 ðPt0 � PðtÞÞdt Pt0ðt1 � t0Þ

(1)

The smaller RU is, the less the system loss is, and vice versa. In the recovery stage, the loss of system performance is:

RE ¼

R t2 t1 ðPt0 � PðtÞÞdt

ðPt0 � Pt1Þðt2 � t1Þ (2)

The smaller the RE is, the smaller the system loss is, and vice versa. From the overall perspective, the whole-process system loss RI is:

RI ¼

R t2 t0 ðPt0 � PðtÞÞdt Pt0ðt2 � t0Þ

(3)

The smaller the value of RI is, the smaller system loss is. To sum up, the system resilience index R is defined as:

R¼1 � RI (4)

The larger the value of R is, the stronger the system resilience is, and vice versa.

3. Structural analysis of China’s oil import system

3.1. System boundary

The system boundary needs to be clearly demarcated before estab- lishing the model of China’s oil import system under external shocks (Wang et al., 2017). Since system degradation and recovery are mainly considered in the model, system boundaries are constructed according to oil import, production, consumption, recovery strategies (reserves, coal-to-oil, substitution and compressed demand), etc., as shown in Fig. 2.

3.2. Sub-module analysis

According to the system boundary, China’s oil import system model can be divided into four sub-modules: supply sub-module, reserve sup- ply sub-module, coal-to-oil super-capacity sub-module and demand compression sub-module. By analyzing the boundary and main variables of each sub-module, a causal relationship diagram is drawn and shown in Fig. 3.

i) Supply sub-module. According to the main research content of this paper, in the supply sub-module, the influence of key factors such as import risk, imported crude oil and refined oil, produced crude oil, and conversion coefficient between crude oil and refined oil on the supply rate of refined oil is mainly considered. It includes a main causal path: external risk → rate of oil import → rate of refined oil production → rate of refined oil supply → de- mand satisfaction rate.

ii) Reserve supply sub-module. Generally speaking, oil reserves include strategic reserves and commercial reserves (Gao et al., 2018). The reserve supply studied in this paper only refers to strategic reserve. Since crude oil cannot be directly used in emergencies, the process of converting crude oil into refined oil and transporting it to the destination is mainly studied in the reserve supply sub-module. It includes a main negative feedback loop: refined oil production rate → refined oil production → refined oil transportation rate → refined oil arrival rate → supply rate → demand satisfaction rate → refined oil production rate.

iii) Coal-to-oil super-capacity sub-module. Relevant government departments can take measures such as mobilization mechanisms to release the production potential of enterprises and increase their production capacity when there is an emergency shortage of refined oil. Since China is a large coal-producing country (Wang and Li, 2017), the extraordinary production studied in this paper refers to the extraordinary production of refined oil, excluding the ordinary production of coal. It includes a main negative feedback loop: coal-to-oil production rate → refined oil produc- tion rate → refined oil transportation rate → refined oil arrival rate → supply rate → demand satisfaction rate → coal-to-oil production rate.

iv) Demand compression sub-module. To increase the amount of available refined oil and guarantee the oil supply, it is necessary to take mobilization measures when necessary to actively guide the public or enterprises to reduce the amount of refined oil used Fig. 1. Resilience evolution curve.

S. Chen et al.

Energy Policy 146 (2020) 111795

4

Fig. 2. Boundary of China’s oil import system.

Fig. 3. Causality diagram of China’s oil import system.

S. Chen et al.

Energy Policy 146 (2020) 111795

5

so as to increase the amount of refined oil supplied. The demand compression studied in this paper mainly refers to the reduction of refined oil for civil consumption. Possible measures include: using other energy sources such as electric energy, hydrogen energy, biomass energy and geothermal energy to replace oil, or voluntarily reducing the use, such as using buses, public bicycles and other ways to reduce the consumption of oil. The decrease in the consumption rate in this sub-module is mainly influenced by the voluntary participation of the public, the proportion of energy substitution and demand reduction. The consumption rate reduced by compressing demand includes two feedback loops, one is: publicity intensity → public participation degree → energy substitution ratio → energy substitution compression quantity → demand compression capacity → consumption rate → demand satisfaction rate → publicity intensity; Another is: publicity in- tensity → public participation degree → proportion of demand reduction → quantity of demand reduction → demand compres- sion capacity → consumption rate → demand satisfaction rate → publicity intensity.

3.3. System flow diagram and equations

The key to the model of China’s oil import resilience under external shocks is to determine the quantitative relationship among variables. According to the causality diagram in Fig. 3, methods such as econo- metric analysis and statistical analysis are used to determine the inter- relationship between variables. In addition, VENSIM software was used to draw a system flow diagram, see Fig. 4.

i) Supply sub-module. Refined oil 1 (RO1) is determined by production rate (PR). Production rate (PR) is determined by domestic production

of crude oil (DPCO), conversion coefficient of crude oil (CCCO) and imported oil rate (IR). Function equations are as follows:

RO1:K ¼RO1:J þ PR:JK*DT (5)

PR¼ðIR * 0:85þDPCOÞ= CCCO (6)

Where, RO1:K represents the value of refined oil 1 at time K; RO1:J represents the value of refined oil 1 at time J; PR:JK represents the value of PR in time interval JK; DT ¼ 1.

Imported oil (IO) is determined by imported oil rate (IR). Imported oil rate (IR) is determined by delivery ratio (DR), transport loss ratio (TLR), system robustness (SR) and disruption switch (DS). Function equations are as follows:

IO:K ¼ IO:J þ IR:JK*DT (7)

​ IR¼ IF THEN ELSEðDS¼0; 140 ; ðDR * 140Þ* TLRþSRÞ (8)

DS¼ IF THEN ELSE ðTime < ¼15; 0; IF THEN ELSE ð Time > ¼15

: AND : Time < ¼85 ; 1 ; IF THEN ELSEðTime > ¼85; 0 ; 1 Þ Þ Þ (9)

Where, the number in Eq. (9) indicates the start and end time of the shock, that is: the length of the shock time, which can be changed ac- cording to the scenario. In addition, System robustness refers to the ability to withstand external shocks during the degradation of the sys- tem, and its value is assumed to be 5 in this paper.

ii) Reserve supply sub-module. Crude oil in reserve (COR) is determined by crude oil conversion rate (COCR). Crude oil conversion rate (COCR) is determined by crude oil in reserve (COR), conversion delay time (CDT), maximum capacity (MC) and demand satisfaction rate (DSR). Function equations are as follows:

Fig. 4. Flow chart of China’s oil import system under external shocks.

S. Chen et al.

Energy Policy 146 (2020) 111795

6

COR:K ¼COR:J � COCR:JK*DT (10)

COCR¼DELAYð IF THEN ELSEðDSR < 1; IF THEN ELSEðCOR > 100; MINðMC; ð1 � DSRÞ* 157 * 30Þ ; 0 Þ; 0 Þ ; CDT ; 0 Þ (11)

Among them: In order to prevent the reserve of crude oil from falling to a negative value, the number 100 (assumed to be X1) in Eq. (11) is set according to the conversion delay time (CDT).

X1¼CDT*MINðMC; ð1 � DSRÞ* 157 * 30Þ (12)

The number 30 (assumed to be X2) in Eq. (11) refers to the time length of shock, which is consistent with the length of DS.

Refined oil 2 (RO2) is determined by oil transportation rate1 (TR1) and crude oil conversion rate (COCR). Oil transportation rate1 (TR1) is determined by refined oil 2 (RO2), transport oil capacity (TOC), safety factor of transporting oil (SFTO) and the company’s ability to secure the transportation of oil (CASTO). Function equations are as follows:

RO2:K ¼R02:J þðTR1:JK � COCR:JKÞ*DT (13)

TR1¼ IF THEN ELSEðRO2 > 0; TOC * SFTO ; 0 Þ (14)

SFTO¼0:1* CASTO (15)

Stock of refined oil in transit (SROT) is determined by oil trans- portation rate1 (TR1) and arrival rate1 (AR1). Arrival of Reserve Transfer (ART) is determined by Arrival rate1 (AR1). Arrival rate1 (AR1) is determined by transport delay time (TDT), Stock of refined oil in transit (SROT) and oil transportation rate1 (TR1). Function equations are as follows:

SROT:K ¼SROT:J þðAR1:JK � TR1:JKÞ*DT (16)

AR1¼DELAYð IF THEN ELSEðSROT > 0; TR1 ; 0 Þ ; TDT Þ (17)

ART:K ¼ART:J þ AR1:JK*DT (18)

iii) Coal-to-oil super-capacity sub-module. Coal for oil production (COP) is determined by Coal-to-oil productivity (CTOP). Coal-to- oil productivity (CTOP) is determined by Coal for oil production (COP), production delay time (PDT), demand satisfaction rate (DSR) and extraordinary capacity (EC). Function equations are as follows:

COP:K ¼COP:J � CTOP:JK*DT (19)

CTOP¼DELAY ðIF THEN ELSEðDSR < 1; IF THEN ELSEðCOP > ¼ 150; MINðEC; ð1 � DSRÞ* 157 * 70Þ ; 0 Þ; 0 Þ ; PDT; 0 Þ (20)

Among them: In order to avoid that the amount of coal used for oil production falls to a negative value, the number 150 (assuming X3) in Eq. (20) is set according to the production delay time (PDT).

X3¼PDT*MINðEC;ð1 � DSRÞ* 157 * 70Þ (21)

The number 70 in Eq. (20) (assuming X4) refers to the duration of the shock, which is consistent with the duration of DS (same as the setting in Eq. (11)).

Refined oil 3 (RO3) is determined by coal-to-oil productivity (CTOP) and oil transportation rate2 (TR2). Transportation rate2 (TR2) is determined by Refined oil 3 (RO3), transport oil capacity1 (TOC1), safety factor of transporting oil 1(SFTO1) and the company’s ability to secure the transportation of oil1 (CASTO1). Function equations are as follows:

RO3:K ¼RO3:J þðTR2:JK � CTOP:JKÞ*DT (22)

TR2¼ IF THEN ELSEðRO3 > 0; TOC1 * SFTO ; 0 Þ (23)

SFTO1¼0:1*CASTO1 (24)

Stock of coal-to-oil in transit (SCOT) is determined by oil trans- portation rate2 (TR2) and arrival rate2 (AR2). Arrival rate2 (AR2) is determined by transport delay time 1 (TDT1), stock of coal-to-oil in transit (SCOT) and transportation rate2 (TR2). Function equations are as follows:

SCOT:K ¼SCOT:J þðAR2:JK � TR2:JKÞ*DT (25)

AR2¼DELAYð IF THEN ELSEðSCOT > 0; TR2 ; 0 Þ ; TDT 1 Þ (26)

ACO:K ¼ACO:J þ AR2:JK*DT (27)

iv) Demand compression sub-module. Compressed consumption (CC) is an integral function of demand compression capability (DCC). The compression consumption delay time is the response delay time (RDT). The demand compression capability (DCC) is the first order delay function of the energy alternative compres- sion (EAC) and the amount of demand compression (DC). The proportion of energy substitution and the proportion of demand reduction are determined by the degree of public participation (DPP) and the degree of Mobilization mechanism perfection (MMP). According to the substitution ratio of electric energy, hydrogen energy and biomass energy to oil, we assume that the maximum energy substitution ratio (M1) is 0.05, and the maximum energy substitution ratio (M2) of residents voluntarily giving up using oil is 0.05 in this paper. The degree of public participation (DPP) is influenced by the degree of publicity (DP), and there is a certain delay in the public participation, which is the Publicity delay time (PDT). The degree of publicity (DP) is influenced by the demand satisfaction rate (DSR). Function equations are as follows:

CC:K ¼CC:J þ DCC:JK*DT (28)

DCC¼DELAYðEACþDC; RDTÞ (29)

EAC ¼C*M1 (30)

DC ¼C*M2 (31)

M1¼MINðDPP * MMP; 0:1Þ (32)

M2¼MINðDPP * MMP:0:05Þ (33)

DPP¼DELAYð0:8 * DP ; PDTÞ (34)

P¼ð1 � DSRÞ*0:9 (35)

Where, the range of DPP and MMP is (0, 1], and parameter C refers to the normal consumption of residents.

v) System resilience. System resilience (SR) is determined by shadow area (SA). The shaded area refers to the area bounded by the “resilience evolution curve” and the time axis. Shadow area (SA) is determined by rate (R). Rate (R) is determined by demand satisfac- tion rate (DSR). Function equations are as follows:

SR¼SA=T (36)

SA:K ¼SA:J þ SA:JK þ R*DT (37)

R¼ IF THEN ELSEð DSR; 0 ; DSRÞ (38)

Where, T is the duration of external impact.

S. Chen et al.

Energy Policy 146 (2020) 111795

7

4. Simulation on resilience of China’s oil import system under external shocks

China’s oil import system model under the external shocks is con- structed in section 3. To verify the feasibility and effectiveness of the model in various scenarios and calculate the system resilience, the model is simulated and analyzed according to different parameters in this section.

4.1. Parameter setting and data source

Parameter setting in this research is mainly based on the supply and demand of China’s crude oil and refined oil in 2017. Refined oil in this paper refers to gasoline, kerosene and diesel fuel that can be directly used for emergency supplies. Exogenous variables of the model are shown in Table 1. In addition, the initial time, end time and step length of the model shown as follows: Initial time ¼ 1, Final time ¼ 160, Time step ¼ 1 Day, Unit of time: Day.

The data in Table 1 mainly comes from: “China Statistical Yearbook” (NBS, 2019a,b), “China Energy Statistical Yearbook” (NBS, 2019a,b), “China National Petroleum Corporation Yearbook” (CNPC, 2019), In- ternational Trade Center (ITC), China National Petroleum News Center (http://news.cnpc.com.cn/), relevant references and inferences (Liu et al., 2015; Gong and Zhang, 2017; Wang and Qu, 2019; Wang and Zhang, 2019). For the obtained annual data, the representative param- eter values in terms of days and tons/day were obtained by means of arithmetic average and weighted average in this paper. For some missing data, it was deduced by means of median filling, regression filling and expert estimation.

4.2. Model test

4.2.1. Test on extreme case Assuming that no external shocks occur, the changes on arrival of

reserve transfer, arrival of coal to oil, compressed consumption and demand satisfaction rate in the model were observed. The results are shown in Fig. 5. When the disruption switch ¼ 0, the arrival of reserve transfer ¼ 0, the arrival of coal to oil ¼ 0, the compressed consumption ¼ 0, and the demand satisfaction rate ¼ 1, which is consistent with the reality, other variables have also been tested in this paper under extreme case, and all have passed the test, due to limited space, the rest are not discussed.

4.2.2. Reality test To verify whether the model follows the basic law of conservation of

material, whether the conservation of material is followed between

crude oil in reserve (raw material) and arrival of reserve transfer (finished product), coal for oil production (raw material) and arrival of coal to oil (finished product) were tested after running the model. Since the conversion rate of coal-to-oil is 46%, and the conversion coefficient of crude oil and refined oil is 1.6 (Wang and Zhang, 2019), in the model, the amount of raw materials is converted to the amount of finished products. That is, through the conversion, raw materials: finished products ¼ 1:1. In this way, the model can be simplified. Therefore, the amount of raw materials and the finished product should be conserved. The results are shown in Fig. 6. The raw material and the finished product are conserved, which is consistent with the reality, indicating that the model passes the reality test.

4.2.3. Sensitivity test Sensitivity test is mainly used for estimating some parameters in the

model, or to testing some structures in the system when they are not very accurate. The purpose is to test the sensitivity of model operation results by changing the value of parameter (Li, 2009; Zhong et al., 2013). Assuming that the company’s ability to secure the transportation of oil (CASTO) follows a random and uniform distribution between [6–10], the sensitivity of arrival of coal to oil (ACO) to the change of CASTO is tested after the model is run for 200 times, and the results are shown in Fig. 7.

As can be seen from Fig. 7, due to the change of CASTO, the value of ACO varies within a certain range and keeps the same trend over time, without over-sensitivity or insensitivity. Additionally, the sensitivity test of other variables was also carried out in this paper, and the results were the same as above. Due to limited space, the rest are not discussed.

4.3. Analysis of system resilience under different measures

According to the model structure and parameters mentioned above, in this section, the system resilience values under different measures are simulated under the same external shocks. Model time: 160 days, duration of external shocks: from day 15 to day 45, delivery ratio ¼ 0.7, transportation loss ratio ¼ 0.2, simulation results are shown as follows:

4.3.1. Scenario 1: No measure As shown in Fig. 8, when external shocks occur, if no measures were

taken, the arrival of reserve transfer ¼ 0, the arrival of coal to oil ¼ 0, the compressed consumption ¼ 0, The demand satisfaction rate drops to the lowest during the period of the external shocks, and there is no sign of recovery until the external shocks disappears. According to the resil- ience measurement proposed in section 2, the system resilience value in scenario 1 is 0.754, and it will continue to decline with the increased degree of external shocks. In extreme cases, if the external shocks degree is too strong, the demand satisfaction rate may directly drop to 0, and will not rise again, let alone restore to the original level.

4.3.2. Scenario 2: one measure As shown in Fig. 9, when external shocks occur, if one certain

measure is taken, the demand satisfaction rate shows signs of recovery, but the degree of recovery varies with different measures. Among them, the effects of the measures from the strong to the weak are: measure 1 (reserve supply), measure 2 (coal to oil supply), measure 3 (demand compression). According to the resilience measurement proposed in section 2, resilience values of (a), (b) and (c) in Fig. 9 are 0.807, 0.780 and 0.756, which are all higher than the resilience values without any measures taken (0.754), indicating that the implementation of the measurement has enhanced the resilience of China’s oil import system effectively.

4.3.3. Scenario 3: diversity measures As shown in Fig. 10, when external shocks occur, three measures

(reserve supply, coal to oil supply, demand compression) are taken simultaneously, the demand satisfaction rate shows obvious signs of

Table 1 Parameters of China’s oil import system under external shocks.

Type variable Value Units

Level Crude oil in reserve 3200 10kt Coal for oil production 6521 10kt

Initial Available refined oil 0 10kt Constant Domestic production of crude oil 56 10kt/day

Conversion coefficient of crude oil 1.6 none Export 14 10kt/day Inventory balance or other 40.625 10kt/day Conversion delay time 2 day Maximum capacity 50 10kt/day Transport oil capacity 22.22 10kt/day Safety factor of transporting oil 9 none Transport delay time 2 none Extraordinary capacity 30 10kt/day Production delay time 5 day Transport delay time 2 2 day Publicity delay time 10 day Mobilization mechanism perfection 0.8 none Response delay time 15 day

S. Chen et al.

Energy Policy 146 (2020) 111795

8

recovery, and the degree of recovery is stronger than the implementa- tion of one measure as shown in Fig. 9. According to the resilience measurement proposed in section 2, the system resilience value in sce- nario 3 is 0.8363, which are all higher than the resilience values of scenario 1 and scenario 2, indicating that the implementation of diver- sification measures is more easier to enhance the resilience of China’s oil import system.

4.4. Contribution of different measures to system recovery

The implementation of various measures is crucial to system recov- ery when the system is exposed to external shocks. However, under different external shocks, the various measures contribute differently to the system recovery. Therefore, the following four external shocks sce- narios are set and simulated to help policy makers optimize different combinations of measures and improve system resilience. The external shocks intensity was set from weak to strong, and the simulation results are shown as follows:

Scenario 4 (DS1): duration of external impact: from 15-85day,

delivery ratio: 0.7, transportation loss ratio: 0.2, duration of model: 300 days. As shown in Fig. 11, the total supply of various measures increased rapidly on days 15–84, reaching 26.6 million tons on day 84. During this period, the total supply mainly comes from the arrival of reserve transfer and coal-to-oil, which are 15 million tons and 8.8 million tons, respec- tively. The compressed consumption is relatively small, which is 2.8 million tons. On the 85th day, the growth rate of the total supply reached an inflection point, which gradually slowed down and became stable. This was mainly because the deviation between the supply and demand of refined oil was slowly reduced and the demand satisfaction rate was gradually increased. Therefore, it was unnecessary to continue to transporting the crude oil reserve and the extraordinary production of coal-to-oil.

Scenario 5 (DS2): duration of external impact: from 15-120day, de- livery ratio: 0.5, transportation loss ratio: 0.5, duration of model: 300 days. As shown in Fig. 12, the total supply of various measures increased rapidly on days 15–89, reaching 37.6 million tons on day 89. During this period, the total supply mainly comes from the arrival of reserve transfer and coal-to-oil, which are 20 million tons and 12.8 million tons,

Fig. 5. Test results under extreme case.

Fig. 6. Results of reality test.

S. Chen et al.

Energy Policy 146 (2020) 111795

9

respectively. The compressed consumption is relatively small, which is 4.8 million tons. On the 90th day, the growth rate of the total supply reached an inflection point, which gradually slowed down. This is due to the fact that the crude oil reserves have been used up, from the 90th to the 120th day, the total supply is obtained from the extraordinary pro- duction of coal-to-oil and the compressed consumption.

Scenario 6 (DS3): duration of external impact: from 15-150day, de- livery ratio: 0.5, transportation loss ratio: 0.5, duration of model: 300 days. As shown in Fig. 13, it is similar to situation 5 for the first 90 days. The total supply of various measures increased rapidly on days 15–89, reaching 37.6 million tons on day 89. During this period, the total supply mainly comes from the arrival of reserve transfer and coal-to-oil, which are 20 million tons and 12.8 million tons, respectively. The compressed consumption is relatively small, which is 4.8 million tons. On the 90th day, the growth rate of the total supply reached an inflection point, which gradually slowed down. This is due to the fact that the crude oil reserves have been used up. In order to continue to replenish the supply, the total supply in the following 90–150 days is obtained from the

extraordinary production of coal-to-oil and the compressed consump- tion. Due to the long duration of external shocks, the arrival of coal-to- oil continued to rise, exceeding the maximum arrival of reserve transfer.

Scenario 7 (DS4): duration of external impact, from 15-160day, de- livery ratio: 0.5, transportation loss ratio: 0.5, duration of model: 300 days. As shown in Fig. 14, it is similar to situation 5 for the first 90 days. On the 90th day, the growth rate of the total supply reached an inflection point, which gradually slowed down. This is due to the fact that the crude oil reserves have been used up. In order to continue to replenish the supply, the total supply in the following 90–150 days is obtained from the extraordinary production of coal-to-oil and the compressed consumption. However, after 156 days, due to the limited arrival of coal- to-oil, the subsequent supply is mainly obtained from compressed consumption.

In summary, when the degree of external shock is relatively low, the contribution of crude oil reserve to the demand satisfaction rate is the largest. With the increase of the external shock degree, the contribution of the extraordinary production of coal-to-oil and the compressed

Fig. 7. The sensitivity test results of ACO.

Fig. 8. Simulation results of scenario 1.

S. Chen et al.

Energy Policy 146 (2020) 111795

10

Fig. 9. Simulation results of scenario 2.

S. Chen et al.

Energy Policy 146 (2020) 111795

11

consumption to the demand satisfaction rate gradually increases. Therefore, in the event of catastrophic disasters, it is often difficult to meet the emergency needs with strategic supplies only from reserves. Therefore, it is necessary to mobilize enterprises and the public to actively participate in production and reduce the demand to realize effective supply.

4.5. Influence of main factors on system resilience

To assess the impact of the changes of various parameters in the model on the system resilience, the external shock parameters were uniformly set as follows: duration of external impact: 15–120 days, delivery ratio: 0.5, transportation loss ratio: 0.5, and duration of: 300 days. Through literature reference and expert consultation, 4 represen- tative parameters which may have important influence on system resilience were selected for the research.

4.5.1. Influence of crude oil in reserve (COR) on system resilience There is only crude oil is stored China’s strategic oil reserves. In

2017, China’s strategic oil reserves were 37.73 million tons. According to the National Oil Reserve Construction Plan, China will achieve the target of 85 million tons of strategic oil reserves in 2020. Therefore, based on China’s current oil reserves and the target to be achieved in 2020, the value range of crude oil reserves is set at [3000–9000]. The conversion coefficient of crude oil to refined oil is 1.6, so the range of conversion to refined oil is [1875–5625]. Limited to the clarity of the curve (the value gap is too small, the curve will be close to coincide), the initial values of COR are set to be 1500, 1900, 2300, 3000, and the corresponding demand satisfaction rate (DSR) curve is shown in Fig. 15. Moreover, the initial values of COR are set to 1500, 1700, 1900, 2100, 2300, 2500, 2700, 3000, and the corresponding system resilience values are shown in Fig. 16.

According to Fig. 15, DSR shows different changes after the 70th day,

Fig. 10. Simulation results of scenario 3.

Fig. 11. Simulation results of scenario 4.

S. Chen et al.

Energy Policy 146 (2020) 111795

12

the greater the COR is, the better the DSR is. Similarly, according to Fig. 16, the larger COR is, the higher the value of system resilience, which indicates that COR is positively correlated with system resilience. Therefore, China should focus on strengthening the construction of na- tional crude oil reserve capacity, so as to improve the resilience im- ported oil system in China.

4.5.2. Influence of conversion coefficient of crude oil (CCCO) on system resilience

According to relevant data of SINOPEC, one ton of crude oil in SINOPEC can be refined into 0.177 tons of gasoline and 0.383 tons of diesels, with a rate of 60.8%. According to relevant data of Petro-China, one ton of crude oil in Petro-China can be refined into 0.215 tons of gasoline and 0.394 tons of diesels, with a rate of 60.8%. Therefore, every ton of refined oil needs 1.6 tons of crude oil on average in China, while the refining technology of developed countries in Europe and the United States is relatively advanced, and the yield rate is about 80%. That is to

say, every ton of refined oil needs 1.25 tons of crude oil on average. Based on this, the value range of CCCO is set to [1.3–1.6], and the curve of DSR obtained is shown in Fig. 17.

As can be seen from Fig. 17, the greater the CCCO, the greater decline degree of DSR. Similarly, the value of system resilience decline with the increase of CCCO. When CCCO is 1.3, 1.4, 1.5, 1.6, the corresponding system resilience values are: 0.777, 0.748, 0.724, 0.702 respectively, which shows that there is a negative correlation between CCCO and system resilience. The yield of refined oil in China is only 63%, signifi- cantly lower than that in developed countries, which is 80%. Therefore, China should pay more attention to improving the refining efficiency of refined oil in Chinese enterprises.

4.5.3. Influence of company’s ability to secure the transportation of oil (CASTO) on system resilience

The increase of the transfer rate depends on the maximum transfer capacity and the transfer safety coefficient. However, it is unrealistic to

Fig. 12. Simulation results of scenario 5.

Fig. 13. Simulation results of scenario 6.

S. Chen et al.

Energy Policy 146 (2020) 111795

13

increase the maximum transfer capacity in a short time after the external shocks. The transfer safety coefficient depends on the CASTO, Therefore, CASTO is selected as a policy parameter for simulation. By changing the value of parameters, the influence of transfer rate on demand satisfac- tion rate and system resilience was simulated. CASTO is gradually increased from 6 to 10, indicating that the company’s ability to guar- antee the oil safety is gradually improved until it can fully guarantee the oil safety. The simulation results are shown in Fig. 18.

As can be seen from Fig. 18, DSR increases with the increase of CASTO. Similarly, the value of system resilience increase with the in- crease of CASTO. When CASTO is 6, 7, 8, 9, 10, the corresponding sys- tem resilience values are: 0.672, 0.684, 0.694, 0.702 and 0.709. There are two transport rates in the model. Similarly, the influence of transport rate 1 on demand satisfaction rate and system resilience also can be obtained. The above conclusions show that by increasing manpower, material resources and financial resources, CASTO and the amount of transported refined oil can be increased, which will also reduce the

decrease of demand satisfaction rate and enhance the system resilience.

4.5.4. Influence of energy substitution ratio (ESR) on system resilience In order to build a clean, low-carbon, safe and efficient modern en-

ergy system, accelerating the development of alternative energy is an inevitable choice for the adjustment and optimization of the energy structure (Tan et al., 2019). In recent years, China has witnessed rapid development of alternative energy, ranking among the top or first in the world in terms of total development, new capacity, new investment and share of consumption. According to the replacement rate of new energy vehicles, biomass fuels and coal-based liquid fuels to oil in 2020, 2030 and 2050, the range of energy substitution ratio is set [5%–35%](Niu et al., 2017), indicating that the ratio of energy substitution is gradually increasing. The simulation results are shown in Fig. 19.

According to Fig. 19, by gradually increasing the ESR, the DSR is gradually increasing. Similarly, the system resilience is also gradually improving. Specifically, when the ESR increased from 5% to 35%, the

Fig. 14. Simulation results of scenario 7.

Fig. 15. Influence of COR on DSR

S. Chen et al.

Energy Policy 146 (2020) 111795

14

system resilience values are 0.702, 0.715, 0.724 and 0.731, respectively. The above conclusions show that the ESR is positively correlated with the DSR and system resilience. Therefore, Chinese petroleum enterprises should actively adjust their development strategies, optimize their business segments and distribution, and enhance their innovation ca- pabilities in order to achieve sustainable development.

According to the simulation results of the above 4 parameters, it is found that the single variable has influence on the system resilience, but it is limited. However, if the factors positively correlated with the system resilience are comprehensively enhanced and the factors negatively correlated with the system resilience are reduced, the system resilience will be greatly increased.

4.6. The bottom line of China’s oil import system under external shocks

Resilience researches aim to enhance the ability of systems to respond to risks. In the above simulation analysis, we discussed the

influencing factors of resilience from various aspects, and identified the direction of enhancing system resilience. In this section, the following questions are discussed: to what extent can China’s oil import system withstand external shocks? Can the system be restored autonomously before the external shock is over? As an important part of the risk management system, this kind of bottom-line thinking can help policy makers estimate the worst-case scenario, increase the vigilance of oil import risks, and thus seize the initiative, improve the risk management system and enhance the ability to prevent risks.

The degree of external shock includes the length of shock time and the strength of shock degree. Therefore, the following four scenarios are set according to different parameters: Weak shock degree and short shock time (DR ¼ 0.8 TLR ¼ 0.1 t ¼ 30), weak shock degree but long shock time (DR ¼ 0.8 TLR ¼ 0.2 t ¼ 275), strong shock degree but short shock time (DR ¼ 0.5 TLR ¼ 0.5 t ¼ 30), strong impact degree and long impact time (DR ¼ 0.4 TLR ¼ 0.5 t ¼ 30), which were respectively represented by S1, S2, S3 and S4. The simulation results are shown in

Fig. 16. Influence of COR on system resilience.

Fig. 17. Influence of CCCO on DSR

S. Chen et al.

Energy Policy 146 (2020) 111795

15

Fig. 20. According to the comparison of curves 1 and 2 in Fig. 20, it is found

that when the degree of external shock is relatively weak, the system has the ability to recover autonomously, but if the time is too long, the system performance will decline again after recovery. Therefore, even if the degree of external shock is weak, if the impact time lasts too long, the system performance will eventually fail to return to its original level. Comparing the curves 3 and 4, it is found that in the case of a strong shock, the system recovery ultimately needs to rely on the recovery of external imports. The simulations in sections 4.3-4.5 all belong to the case of curve 3, indicating that the imports has been restored in time, but if the external shock continues, the system performance will decline after the improvement, and it will not recover until the external shock stops, which indicates that if the external shock degree is too strong, the system cannot restore autonomously before the external shock ends, and vice versa. Therefore, whether the degree of external shock is too strong or the time is too long, it will cause the system to be unable to recover autonomously.

So what extent and how long is the threshold that the system can still recover? Due to the complexity of external shocks, there are countless combinations of the degree and time of the impact, so we can first determine the length of time, and then find the bottom line of the shock degree that can withstand, and vice versa. Specifically, taking the time of external shock from 15th to 120th as an example, the bottom line of the shock degree that China’s oil import system can withstand is explored. As shown in Fig. 21, when the shock degree is less than or equal to 667,800 tons/day (DR ¼ 0.55 TLR ¼ 0.1), the system can recover autonomously; otherwise, it needs to rely on external imports for recovery. In summary, the prerequisite to judge whether the system can be recovered and how long it can be recovered is to estimate the parameters of external shocks. Policy makers can roughly determine the parameters of external shocks according to the occurrence of events, so as to determine the bottom line of external shocks that the system can bear and formulate effective policies in a timely manner.

Here, the simulation has been completely completed. We not only simulated the influence factors, but also successfully calculated the

Fig. 18. Influence of CASTO on DSR

Fig. 19. Influence of ESR on DSR

S. Chen et al.

Energy Policy 146 (2020) 111795

16

system resilience values under different scenarios. As described in sec- tion 1, this is extremely rare in previous literature. Moreover, this model can also be applied to other scenarios, such as the measurement of the natural gas and coal systems.

5. Conclusion and policy implications

Based on the resilience evolution curve, through the SD model, this paper depicts the resilience curve of the system, calculates the resilience value under different scenarios, analyzes the key influencing factors of the system, and identifies the threshold of the system according to the specific scenario, which provide theoretical basis for the normal oil import system. The research results show that: i) Diversified measures can conduce to the system recover; ii) When the system is faced with external shocks, the strategic reserve of crude oil contributes the most to the recovery of the system, followed by the extraordinary production of coal-to-oil, and the supplementary measure is to reduce demand; iii) Through increasing COR,CASTO,ESR, and reducing CCCO, the rapid

replenishment of oil can be realized under external shocks; iv) There is a threshold for totally failure in the system. Too strong shock strength and too long shock time both have a dead impact on the system that cannot recover autonomously. Therefore, the threshold of the system should be estimated according to the external shocks to reduce the economic loss. Based on the above conclusions, the following suggestions are proposed:

(1) Increase diversity of measures. Maintaining the performance of China’s oil import system cannot only depend on the strategic reserves of crude oil. The extraordinary production and the cooperation of the general publics are equally important. (2) In- crease strategic reserves of crude oil. Strategic crude oil reserve plays an obvious role in maintaining the DSR of refined oil when external shocks occur. Therefore, a crude oil reserve mobilization center can be set up to store crude oil in the form of enterprises, so as to realize the rapid supply of crude oil under dangerous con- ditions. (3) Increase the production efficiency of refined oil. Due to the relatively backward industrial development in China, the

Fig. 20. Simulation on system bottom line.

Fig. 21. Simulation on system bottom line under a particular scenario.

S. Chen et al.

Energy Policy 146 (2020) 111795

17

product conversion rate is relatively low, resulting in the waste of crude oil. (4) Increase the proportion of energy substitutes. Accelerating the development of alternative energy is an inevi- table choice for the adjustment and optimization of energy structure in the future. The increase in the proportion of energy substitution can enhance the resilience of the oil import system. (5) Establish a sense of risk and estimate the bottom line of risk. Policy makers should make a realistic assessment of the worst- case scenario by calculating the risks, screen out various poten- tial risks, find the dividing line between safety and risk, and defend the bottom line of various risks when external shocks come.

CRediT authorship contribution statement

Sai Chen: Writing - review & editing, Funding acquisition. Ming Zhang: Project administration, Investigation, Methodology, Supervi- sion. Yueting Ding: Data curation. Rui Nie: Formal analysis.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

The authors gratefully acknowledge the financial support from the “Outstanding Innovation Scholarship for Doctoral Candidate of CUMT” (2019YCBS035). We would like to thank the anonymous referees for their constructive suggestions and valuable comments on the earlier draft of our paper, upon which we have improved the content.

References

Afgan, N., Veziroglu, A., 2012. Sustainable resilience of hydrogen energy system. Int. J. Hydrogen Energy 37, 5461–5467.

Amirioun, M.H., Aminifar, F., Lesani, H., 2017. Resilience-oriented proactive management of microgrids against windstorms. IEEE T. Power Syst. 33, 4275–4284.

Amirioun, M.H., Aminifar, F., Lesani, H., Shahidehpour, M., 2019. Metrics and quantitative framework for assessing microgrid resilience against windstorms. Int. J. Elec. Power 104, 716–723.

Beccue, P.C., Huntington, H.G., Leiby, P.N., Vincent, K.R., 2018. An updated assessment of oil market disruption risks. Energy Pol. 115, 456–469.

Bruneau, M., Chang, S.E., Eguchi, R.T., Lee, G.C., O’Rourke, T.D., Reinhorn, A.M., Shinozuka, M., Tierney, K., Wallace, W.A., Von Winterfeldt, D., 2003. A framework to quantitatively assess and enhance the seismic resilience of communities. Earthq. Spectra 19, 733–752.

Cimellaro, G.P., Solari, D., Bruneau, M., 2014. Physical infrastructure interdependency and regional resilience index after the 2011 Tohoku Earthquake in Japan. Earthq. Eng. Struct. Dynam. 43, 1763–1784.

CNPC, 2019. Yearbook of China National Petroleum Corporation. Petroleum Industry Press.

Erker, S., Stangl, R., Stoeglehner, G., 2017a. Resilience in the light of energy crises–Part I: a framework to conceptualise regional energy resilience. J. Clean. Prod. 164, 420–433.

Erker, S., Stangl, R., Stoeglehner, G., 2017b. Resilience in the light of energy crises–Part II: application of the regional energy resilience assessment. J. Clean. Prod. 164, 495–507.

Gao, D., Li, Z., Liu, P., Zhao, J., Zhang, Y., Li, C., 2018. A coordinated energy security model taking strategic petroleum reserve and alternative fuels into consideration. Energy 145, 171–181.

Garbolino, E., Chery, J.P., Guarnieri, F., 2016. A simplified approach to risk assessment based on system dynamics: an industrial case study. Risk Anal. 36, 16–29.

Gong, L., Zhang, J., 2017. System-dynamics-based emergency transportation strategy analysis for oil products. Syst. Eng. Theor. Pract. 37, 2256–2267.

Guo, Q., 2017. Study on the Evolution Process and Emergency Decision of Sudden Oil Shortage. China University of Mining and Technology.

Henry, D., Ramirez-Marquez, J.E., 2012. Generic metrics and quantitative approaches for system resilience as a function of time. Reliab. Eng. Syst. Saf. 99, 114–122.

ISDRUN, 2004. Living with Risk: a Global Review of Disaster Reduction Initiatives. BioMed Central Ltd, London (London).

Kim, Y., Chen, Y., Linderman, K., 2015. Supply network disruption and resilience: a network structural perspective. J. Oper. Manag. 33, 43–59.

Kulig, J.C., Hanson, L., 1996. Discussion and Expansion of the Concept of Resiliency: Summary of a Think Tank Held at the University of Lethbridge, September 13 & 14, 1996. Regional centre for health promotion and community studies, University of ….

Li, X., 2009. Social System Dynamics Principle, Method and Application of Policy Research. Fudan University Press, Shanghai.

Linkov, I., Eisenberg, D.A., Bates, M.E., Chang, D., Convertino, M., Allen, J.H., Flynn, S. E., Seager, T.P., 2013. Measurable Resilience for Actionable Policy. ACS Publications.

Liu, J., Zhao, S., Liu, X., 2015. Coping strategies and simulation analysis on municipal water supply system resilience. Syst. Eng. Theor. Pract. 35, 2637–2645.

Liu, X., Zeng, M., 2017. Renewable energy investment risk evaluation model based on system dynamics. Renew. Sustain. Energy Rev. 73, 782–788.

Mari, S.I., Lee, Y.H., Memon, M.S., Park, Y.S., Kim, M., 2015. Adaptivity of complex network topologies for designing resilient supply chain networks. Int. J. Ind. Eng. 22.

Matzenberger, J., Hargreaves, N., Raha, D., Dias, P., 2015. A novel approach to assess resilience of energy systems. International Journal of Disaster Resilience in the Built Environment.

NBS, 2019a. China Statistical Yearbook. China Statistics Press, Beijing. NBS, 2019b. China Energy Statistics Yearbook. China Statistics Press, Beijing. Niu, L., Wang, B., Yan, M., 2017. Analysis of the impact of alternative energy

development on China’s oil demand. Petrol. Sci. Technol. Forum 11–19. O Brien, G., Hope, A., 2010. Localism and energy: negotiating approaches to embedding

resilience in energy systems. Energy Pol. 38, 7550–7558. Omer, M., Mostashari, A., Nilchiani, R., 2013. Assessing resilience in a regional road-

based transportation network. Int. J. Ind. Syst. Eng. 13, 389–408. Ouyang, M., 2017. A mathematical framework to optimize resilience of interdependent

critical infrastructure systems under spatially localized attacks. Eur. J. Oper. Res. 262, 1072–1084.

Ouyang, M., Duenas-Osorio, L., 2014. Multi-dimensional hurricane resilience assessment of electric power systems. Struct. Saf. 48, 15–24.

Ouyang, M., Due~nas-Osorio, L., Min, X., 2012. A three-stage resilience analysis framework for urban infrastructure systems. Struct. Saf. 36, 23–31.

Ouyang, M., Wang, Z., 2015. Resilience assessment of interdependent infrastructure systems: with a focus on joint restoration modeling and analysis. Reliab. Eng. Syst. Saf. 141, 74–82.

Roege, P.E., Collier, Z.A., Mancillas, J., McDonagh, J.A., Linkov, I., 2014. Metrics for energy resilience. Energy Pol. 72, 249–256.

Song, Y., Zhang, M., Sun, R., 2019. Using a new aggregated indicator to evaluate China’s energy security. Energy Pol. 132, 167–174.

Sun, M., Gao, C., Shen, B., 2014. Quantifying China’s oil import risks and the impact on the national economy. Energy Pol. 67, 605–611.

Sun, X., Liu, C., Chen, X., Li, J., 2017. Modeling systemic risk of crude oil imports: case of China’s global oil supply chain. Energy 121, 449–465.

Tan, W.J., Zhang, A.N., Cai, W., 2019. A graph-based model to measure structural redundancy for supply chain resilience. Int. J. Prod. Res. 57, 6385–6404.

Wang, D., Ma, G., Song, X., Liu, Y., 2017. Energy price slump and policy response in the coal-chemical industry district: a case study of Ordos with a system dynamics model. Energy Pol. 104, 325–339.

Wang, Q., Li, R., 2017. Decline in China’s coal consumption: an evidence of peak coal or a temporary blip? Energy Pol. 108, 696–701.

Wang, S., Yang, D., Lu, J., 2018. A connectivity reliability-cost approach for path selection in the maritime transportation of China’s crude oil imports. Marit. Pol. Manag. 45, 567–584.

Wang, Y., Qu, Y., 2019. SD simulation of flexibility of imported crude oil supply chain network under environmental disruption. J. Syst. Manag. 28, 983–990.

Wang, Z., Zhang, J., 2019. An evaluation of emergency-materials mobilization potential based on-system dynamics. Syst. Eng. Theor. Pract. 11, 2880–2895.

Wu, X., Ou, L., Hu, G., 2013. On China’s "oil independence" strategy. Reform Econ. Syst. 167–170.

Xu, M., Wang, X., Zhao, L., 2014. Predicted supply chain resilience based on structural evolution against random supply disruptions. Int. J. Syst. Sci.: Oper. Logist. 1, 105–117.

Yang, Y., Li, J., Sun, X., Chen, J., 2014. Measuring external oil supply risk: a modified diversification index with country risk and potential oil exports. Energy 68, 930–938.

Yao, L., Chang, Y., 2014. Energy security in China: a quantitative analysis and policy implications. Energy Pol. 67, 595–604.

Yao, L.A.Y.C., 2015. Shaping China’s energy security: the impact of domestic reforms. Energy Pol. 77, 131–139.

Yao, L.X.S.A., 2018. Conceptualization of energy security in resource-poor economies: the role of the nature of economy. Energy Pol. 114, 394–402.

Zhang, J., 2019. Oil and gas trade between China and countries and regions along the ‘Belt and Road’: a panoramic perspective. Energy Pol. 129, 1111–1120.

Zhang, Y., Liu, H., 2015. The concept of resilience revisited. Int. City Plann. 30, 13–21. Zhong, Y., Jia, X., Qian, Y., 2013. System Dynamics, second ed. Science, Beijing. Zhou, H., Wan, J., Jia, H., 2010. Resilience to natural hazards: a geographic perspective.

Nat. Hazards 53, 21–41.

S. Chen et al.