Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

Journal of Energy Storage 32 (2020) 101753

Contents lists available at ScienceDirect

Journal of Energy Storage


journal homepage: www.elsevier.com/locate/est

A risk assessment framework of seawater pumped hydro storage project in T


China under three typical public-private partnership management modes
Yunna Wua,b, Ting Zhanga,b, , Kaifeng Chenc, Liqi Yib

a
School of Economics and Management, North China Electric Power University, Beijing 102206, China
b
Beijing Key Laboratory of New Energy and Low-Carbon Development, North China Electric Power University, Changping, Beijing 102206, China
c
State Grid Shanghai Municipal Electric Power Company-Shibei Power Supply Company, China

ARTICLE INFO ABSTRACT

Keywords: Along with the rapid development of pumped hybrid storage in China, the shortage of inland resources has
Risk assessment become a serious problem that restricts its further development. Fortunately, seawater pumped hybrid storage
S-PHS project (S-PHS) processes the advantages in saving cost of lower reservoir construction and making full use of coastal
PPP mode resources, which has attracted great attention in China. In view of the characteristics of expanding sources of
Linguistic hesitant fuzzy sets
infrastructure investment funds and improving financial stability of both public and private sectors, varied
Cloud model
Risk response
public-private partnership (PPP) modes are introduced to promote the construction of S-PHS power plant.
However, due to lack of experience, high investment, long payback period, and so forth, S-PHS project still holds
great uncertain risks in the aspects of economy, environment, technology, etc. Therefore, this paper will study on
the risk assessment of S-PHS project under three typical PPP management modes by linguistic hesitant fuzzy sets
based cloud model. Result shows that the risk level of S-PHS project locates between “Middle High” and “High”,
and PPP-S-PHS processes relatively low risk degree, of which the BOT-S-PHS is the lowest. Moreover, corre­
sponding risk response measures for PPP-S-PHS project are proposed for helping risk prevention and further
smooth implementation of PPP-S-PHS project in China.

1. Introduction build ones for better realizing local energy sustainable development
such as Azores [10], Agulhas Bank [11] and Guadeloupe island [12],
For achieving the goal of self-generated power supply, providing etc. Comparing to inland PHS power plant, S-PHS cuts the construction
cost-effective alternative fueling and improving energy-related opera­ and maintenance cost of lower reservoir and makes full use of natural
tional efficiencies, it is imperative to develop renewable energy. Due to resources [13,14]. To investigate China's building capacity, NEA (Na­
the indirect and uncertain characteristics of most renewable resources tional Energy Administration) has published an S-PHS resource census
like wind and solar, electrical energy storage (EES) is becoming in­ of China in 2017 showing that total 238 potential S-PHS sites are
creasingly important in peak shaving and valley filling as well as stable promising including 174 offshore sites and 64 island sites, with a total
energy power providing [1–3], and is hailed globally as a changer in the installed capacity of 42.08 million kW [15]. Up to now, the utilization
direction of reliable low-carbon management and green power grid [4]. of S-PHS has been widely debated and made much breakthrough on
Among numerical energy storage technologies, pumped hybrid storage integration with wind power [8,16,17], optimization application with
is the most mature and cycle efficient energy option with the lowest photovoltaic energy systems [16], technology details regarding to the
annual operation and maintenance cost, which is particularly suitable design, construction and the operation [18], size analysis combining
for promoting the integration of large-scale renewable energy in large with wind-hydro system [19], effects of low seawater PH on energy
and medium-sized power system [5–7]. Besides, for remote country­ storage [20], performance analysis [21] and nation grid-connected
side, isolated islands and offshore area, pumping seawater from sea to management [22].
an artificial reservoir nearby as a kind of new PHS called seawater PHS As it is the first time for China to introduce S-PHS power station,
(S-PHS) is an interesting option for countries with long coast lines and various risks factors are emerging due to lack of experience, different
scarce freshwater resources [8]. After successfully constructing the first local policies, uncertain developing forms, imperfect technologies, long
S-PHS station in Okinawa japan [9], several countries have intended to construction cycle as well as high overall cost. If these risks cannot be


Corresponding author at: School of Economics and Management, North China Electric Power University, Beijing 102206, China.
E-mail address: 18810078795@163.com (T. Zhang).

https://doi.org/10.1016/j.est.2020.101753
Received 14 March 2020; Received in revised form 14 May 2020; Accepted 4 August 2020
Available online 07 September 2020
2352-152X/ © 2020 Elsevier Ltd. All rights reserved.
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Nomenclature LH A set of Sappending the hesitation level


lh(sθ(i)) The possible membership degrees of the element sθ(i) ∈ S
Symbol Explanation to the set LH
ESS Electrical energy storage E(LH) Expectation function of LH
S-PHS Seawater pumped hybrid storage D(LH) Variance function of LH
NEA National Energy Administration |lh(sθ(i))| The count of real numbers in lh(sθ(i))
PPP Public-Private-Partnership |index(LH)| The cardinality of index(LH)
BOO Build-Own-Operate U Quantitative domain of exact numerical representation
BOT Build-Operate-Transfer C The qualitative concept of U
DBFO Design-Build-Finance-Operate x A random implementation of C
RPO Possibility of occurrence μ(x) The certainty degree of x
RCS Consequence severity T A qualitative concept in U
RUC Uncontrollability of risk consequence Ex Expectation
RPM Preventive measures En Entropy
HFSs Hesitant fuzzy sets He Hyper entropy
LHFSs Linguistic hesitant fuzzy sets Ai A cloud
MS Mean score Xmin Underline of Interval
CRCs Critical risk criteria Xmax Upper Line of Interval
Nor. Normalization ALH* A comprehensive cloud
EH Extremely high d(LH1, LH2) The distance measure between LH1 and LH2
H High Sup(Aαi, Aαj) The support for Aαi from Aαj
MH Middle high βi A weighting operator
N Normal T(Aαi) Comprehensive support from Aαi to other criteria
ML Middle low Sims Similarity degree in shape
L Low Simd Similarity degree in distance
EL Extremely low Simc Comprehensive similarity degree
S A linguistic term set

handled correctly, then despite the huge benefits of S-PHS plant itself, diversification of investors, PPP (Public-Private-Partnership) mode are
the project's failure or loss will be even greater. For instance, Kumar, introduced. As a contract to provide public goods, share risk and gen­
Jindal [23] pointed that the financial risk such as changes in interest erate various social capital to actively participate in the provision of
rates or exchange rate were significant factors in PPP based energy public goods and services so as for a win-win situation [28], PPP mode
projects construction. Besides, Li and Wang [24] proposed that en­ has been widely utilized to accelerate the progress of the development
vironment corrosion, payment risk as well as lack of supporting infra­ of renewable energy [29–32]. Recommended to apply on the renewable
structure were the critical risk factors affecting the implementation of energy to motivate enterprises and share the government pressure, PPP
PPP projects in China. What's more, low seawater pH on energy storage arrangements are far broader and their implications are not restricted to
could have different but significant effects on its equipment and en­ the financing domain [33]. Generally, it can be classified into out­
vironment around [25]. Besides, technical risk and improper operation sourcing, concession and divestiture [34]. The outsourcing model has
and management risk were proposed as key drivers in risk assessment not been widely utilized since it cannot absorb social funds or maximize
for renewable energy projects [26,27]. Due to the inadequate con­ the professional advantages of private capital to reduce project life
sideration, even Japan, which has extensive development experience, cycle cost or improve service quality, and may interrupt several func­
still faces the risk of closing its first S-PHS plant because of corrosion tions [35]. For the types of divestiture and concession, BOO (Build-
problems. Own-Operate) and BOT (Build-Operate-Transfer) are the two of the
For rationally allocating risks, improving the efficiency of project most common modes which have been widely applied in a great many
investment, transforming government functions and promoting infrastructure projects [36]. As another concession mode, DBFO

Fig. 1. 4-dimension loop in project risk management.

2
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Table 1 In most cases, the number of linguistic terms for different LHFSs could
Risk criteria system of S-PHS plant. be different, Liao et al. [41] measured the distance between two LHFSs
First-class criteria Symbol Sub-criteria Symbol by adding maximum and minimum linguistic term and relative mem­
bership degrees. But this method will lead to different distance values
Technical risk C1 Construction technology C11 proved by Zhu et al. [38] who improved it by converting LHFSs into
Maintenance risk C12
comprehensive cloud operator. Acting as a quantitative and qualitative
Corrosion risk C13
Economic risk C2 Payment risk C21
uncertainty conversion model, cloud model successfully depicts the
Revenue risk C22 fuzziness and randomness of a qualitative assessment through three
Inflation C23 numerical characteristics and has been widely utilized such as the risk
Change in tax C24 assessment of PPP waste-to-energy [31], site selection of low-speed
Environmental risk C3 Unforeseen weather or geotechnical C31
wind farm [42], the performance of renewable energy project [43] and
conditions
Environmental pollution nearby risk C32 so on [37,44]. Regarding of the protruding advantages of the LHFSs and
Social risk C4 Force majeure C41 cloud model in decision making process, it is meaningful to apply them
Public opposition risk C42 to assess the overall risk degree of S-PHS project under multiple PPP
modes to get the lowest risk management mode and corresponding risk
response measures.
(Design-Build-Finance-Operate) proposed by British Highway Authority The originality of this paper are three holds. (i) Explore the risk
innovates in the use of service procurement policies to replace tradi­ factors of S-PHS project and typical PPP management modes as BOT,
tional capital asset procurement and provide an objective way to pay. DBFO and BOO for the first time through literature review, book study
Thus, for reasonable sharing, three typical PPP modes as BOT, DBFO as well as expert discussion. (ii) Evaluate the risk level of S-PHS project
and BOO are recommended to be applied on S-PHS project in China. in China under three typical PPP management modes for the first time
As China is planning to start to build S-PHS projects recently, it is via the framework of LHFSs based cloud model which can not only
high time to assess the overall risks level of the S-PHS plant, choose an completely express the characteristics of risk value but also process
effective management mode and provide risk response strategies to information in aspects of fuzziness and randomness. (iii) Risk response
reduce economic losses and environmental hazards and further pro­ for the construction, operation and management of S-PHS project under
mote its smooth development in our country. For a whole process of a various management modes are firstly put forward to promote the
project risk management from the perspective of engineering manage­ smooth implementation of S-PHS project in China.
ment, four stages often occur as prophase prevention, risk occurrence, The remainder of this paper is arranged as follows. Section 2
risk control and its consequence as shown in the Fig. 1. Therefore, this identifies the critical criteria of S-PHS project and three typical PPP
paper adopts the 4-dimention method to comprehensively assess pro­ modes, and establishes corresponding risk assessment criteria systems.
ject risk as preventive measures (RPM), possibility of occurrence (RPO), Basic theories of methods utilized in the risk assessment framework are
consequence severity (RCS), uncontrollability of risk consequence introduce in Section 3. Section 4 assesses the current risk level of S-PHS
(RUC) for formulating a suitable and accurate evaluation result. plant under different PPP modes in China. Finally, risk response as well
Because of the ambiguity and limitation of human emotions, it is as conclusions of the whole paper is formulated in the Sections 5 and 6.
difficult for decision makers to express their risk assessment through
crisp values [37]. Fortunately, the linguistic terms as “high”, “slightly
high”, “low”, “slightly low” and some others can be utilized for ex­ 2. Indicators for various scenarios
pressing the decisions from DMs. In addition, DMs usually provide their
information over alternatives with hesitancy or anonymity [38]. Thus, In this section, we firstly identify the critical risk criteria only re­
Torra [39] proposed hesitant fuzzy sets (HFSs) to express the mem­ lated to S-PHS project in the aspects of technology, economy, en­
bership degree of an alternative to a given assessment set. For com­ vironment and society. Then, the significant risk factors of each man­
prehensively expressing the hesitancy, fuzziness and uncertainty of agement pattern are further selected. Following that, the evaluation
DMs, Meng et al. [40] defined the linguistic hesitant fuzzy sets (LHFSs) criteria system about S-PHS project in each management mode is finally
which is utilized in this paper on account of the determination of risk established.
value of PPP S-PHS project containing great uncertainties and fuzziness.

Table 2
Risks summary of PPP mode.
Number Criteria \ Literatures [32] [54] [31] [55] [51] [56] [57] Total

1 Lack of supporting infrastructure √ √ √ 3


2 Delay in project approvals and permits √ 1
3 Payment risk √ √ √ √ √ 5
4 Concession period √ 1
5 Government corruption √ √ 2
6 Legal and policy risk √ √ √ 3
7 Government decision-making risk √ √ 2
8 Government credit risk √ √ √ 3
9 Imperfect supervision system √ 1
10 Lack of PPP experience √ √ √ 3
11 Land acquisition risk √ 1
12 Government intervention √ √ √ 3
13 No clear division of responsibilities and obligations √ 1
14 Contract change risk √ 1

3
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

2.1. S-PHS plant risk identification experience on project modes. The respondents give their opinions on
the risk factors in each PPP mode based on a 9-point scale (where
Up to now, little attention has been attached to the research of S- 1=extremely low; 2=very low; 3=slightly low; 4=low; 5=fair;
PHS risk criteria identification. Thus, for comprehensively measure the 6=slightly high; 7=high; 8=very high; 9=extremely high) from a 4-
risk criteria of S-PHS power plant, we firstly review former literatures dimention perspective. Finally, a total of 200 questionnaires are sent
on PHS risks such as climate conditions [45], political risk, profit risk out on August 2, 2018, with a deadline of September 3, 2018 through E-
[46], technical risk, ecological risk [47] as well as operation and se­ mail. And the feedback from 11 managers and 31 experts are received
curity risk [48]. After studying literatures, an expert decision group with a 21% response rate meeting the norm rate between 20–30% of
with a scale of seven satisfying the following three conditions as (i) most questionnaire surveys [51]. In addition, 52 mangers and 93 ex­
scholars who have published at least one paper in international journals perts did not give their feedback before the deadline, eleven mangers
related to the theme of S-PHS or PPP-S-PHS projects; (ii) Project man­ and two experts did not complete the questionnaires. Thus, we choose
agers who have managed at least one PPP infrastructure construction or the CRCs based on the effective feedback data by MS method as Eq. (1)
PPP PHS project; (iii) workers who have at least three years working [51].
experience in the PPP infrastructure construction or PPP PHS project; is
(f × s )
established to build a unique risk evaluation criteria system of S-PHS MS = , (1 MS 9)
N (1)
project in China [31]. Based on above studies, literature review and
expert's discussion, a holistic criteria system is formulated as shown in where f is the number of occurrences of each rating, s is the criteria
Table 1. score given by respondents, N is the total responses of one criteria.
Some important points proposed in the expert discussion meeting Subsequently, integrating the 4-dimention scoring results by Eq. (2)
can be summarized as follows. As a mechanical energy storage mode, referring [51] as the ‘Impact’ value.
the use of seawater in PHS plant introduces several issues mainly in the
Impact = 4 MSRPM × MSRPO × MSRCS × MSRUC (2)
aspect of technical nature since seawater leakage from upper reservoir
or penstock could cause serious environmental impacts as the closing Finally, rank the risk factors based on the Impact values as shown in
down of the first S-PHS station in Okinawa, Japan [49]. These risks Table 3. The risk factors with normalized impact values ≥ 0.5 are
should be considered, assessed and minimized for the successful op­ considered as CRCs [31].
eration of S-PHS plant. Moreover, low seawater pH has great effect on
energy storage [20]. Some other issues concerning the operation of 2.2.1. BOT mode
mechanical components were emphasized as the corrosion of metal BOT is a typical classification of PPP mode which is a type of
materials by seawater, the adhesion of marine organisms to the wa­ franchise. Under this mode, the private sector is primarily responsible
terways as well as turbine, animals and other biological system around for the financing, design, construction and operation of the project.
the seawater site [50]. Besides, both the hydro power plant and the Upon the expiration of the concession period, the infrastructure shall be
pump station must be sited at an adequate distance from the coastline transferred to the government free of charge or compensation. Besides,
to protect the buildings from the strong sea waves [49]. this mode has a strong attraction for the private sector because it can
help the private sector obtain long-term operating income and the cost
2.2. Risk determination of PPP mode of the project is relatively stable. Moreover, BOT is a way of project
financing based on project credit with the characteristics of limited-
Although there are plenty of advantages of PPP mode as mentioned recourse. Therefore, bank take much more risks than corporate finan­
above, certain risks are still hiding in it, which can be summarized as cing. But because of the limited-recourse, the debt of BOT project is not
shown in the Table 2. Besides, separating research of the management involved in the balance sheet of the shareholders, and the shareholders
mode and construction project is helpful to clearly distinguish risk can raise the construction funds for more projects, so BOT has been
sources. Thus, a mean score (MS) ranking technique is employed to welcomed by equity bidder and widely used [52,53]. Besides above
specifically select the critical risk factors associated to each PPP mode advantages, BOT may conflict with the requirements of the overall
based on the literature review. In view of the 4-dimention risk assess­ planning and environment. What's more, enterprises could transfer the
ment framework, questionnaire survey designing for the modes of BOT, project to the government when the cost of operation and management
DBFO, BOO is utilized to select their critical risk criteria (CRCs) from rises which will make the management of government more difficult.
the fifteen summarized factors obtained by literature review. For From the collected questionnaire data, the CRCs for BOT mode are
comprehensively selecting the CRCs of the three modes, we select an identified according to their impact values as shown in Table 3, where
evaluation committee with a scale of 200 people including 126 experts “Nor.” means normalization.
from universities who have made achievements in project financing Thus, the overall risk criteria system of BOT-S-PHS project can be
model and 74 managers who have three or more years’ management summarized as the Table 4.

Table 3
CRCs selection of BOT mode.
CRCs RPM RPO RCS RUC Impact Nor. Ranking

Payment risk 6.892 7.291 8.110 7.958 7.546 1 1


Government corruption 5.983 8.115 8.242 7.856 7.488 0.966 2
No clear division of responsibilities and obligations 6.873 8.168 7.564 7.125 7.417 0.924 3
Lack of supporting infrastructure 7.498 8.124 6.210 7.900 7.394 0.911 4
Government intervention 6.783 7.156 8.693 6.989 7.369 0.897 5
Legal and policy risk 6.758 8.124 6.853 7.250 7.227 0.814 6
Lack of PPP experience 6.785 7.526 6.812 7.684 7.190 0.793 7
Government decision-making risk 6.014 6.560 8.322 7.813 7.116 0.749 8

4
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Table 4 Table 6
Risk criteria system of BOT-S-PHS project. Risk criteria system of DBFO-S-PHS project.
First-class criteria Symbol Second-class criteria Symbol First-class criteria Symbol Second-class criteria Symbol

Technical risk C1 Construction technology C11 Technical risk C1 Construction technology C11
Maintenance risk C12 Maintenance risk C12
Corrosion risk C13 Corrosion risk C13
Economic risk C2 Payment risk C21 Imperfect supervision system C14
Revenue risk C22 Economic risk C2 Payment risk C21
Inflation C23 Revenue risk C22
Change in tax C24 Inflation C23
Legal and policy risk C25 Change in tax C24
Environment risk C3 Unforeseen weather or geotechnical C31 Legal and policy risk C25
conditions Concession period C26
Environmental pollution nearby risk C32 Government risk C3 Government decision-making risk C31
Government risk C4 Government decision-making risk C41 Environment risk C4 Unforeseen weather or geotechnical C41
Government corruption C42 conditions
Government intervention C43 Environmental pollution nearby risk C42
Social risk C5 Public opposition risk C51 Social risk C5 Public opposition risk C51
Lack of supporting infrastructure C52 No clear division of responsibilities and C52
No clear division of responsibilities and C53 obligations
obligations Lack of supporting infrastructure C53
Lack of PPP experience C54 Lack of PPP experience C54
Force majeure C55 Force majeure C55

2.2.2. DBFO mode


DBFO mode entrusts all the work from design to operation to a
single private business organization and transfers the property back to 2.2.3. BOO mode
the government after a long-term contract (usually 25 to 30 years). It In such contractual arrangements, control and ownership of the
specifies payment, service standards and the performance evaluation in project are always vested in the private sector. For the projects under
detail, and provides an objective way to pay based on performance. In BOO mode, private entities finance, build, own and actually operate an
general, there is only one public sector and one private sector in the infrastructure permanently. For example, Water treatment plants ser­
contract. Although other private sector entities may be involved, the ving parts of South Australia are financed, designed, constructed and
responsibilities performed are generally assigned to the main private operated by private companies to treat raw water provided by the
sector side of the DBFO contract through subcontracting. Compared public sector as clean water, which is then returned to public utilities
with the traditional public works mode, the main spirit of DBFO mode for consumers. The original contract will be signed with the public
is not just the construction of public facilities but to seek an appropriate sector with the stipulation of the restriction clause guaranteeing the
private business organization to provide public services. The public public welfare, and enterprises shall be managed and supervised by the
sector provides contractual agreements, pays certain cost and utilizes government [52, 58]. Since the ownership of privatized PPP projects
the installation while providing core services related to the facility. And belong to the private ownership permanently and it does not have
the private sector only provides ancillary services to the facility. Despite limited recourse, thus, the private sector bears the greatest risk in such
DBFO has many advantages, the risks hiding in it should be deeply PPP projects.
explored to ensure the success of the project. In the DBFO mode, From the collected questionnaire data, the CRCs for BOO mode are
Franchisors must take substantial risks, such as long-term obligations, identified according to their impact values as shown in Table 7.
to improve business quality. Another feature of DBFO projects is the In addition, the risk criteria systems of BOO-S-PHS project can be
involvement of private venture capital, and investors who expect their established through comprehensive analysis of project risk and mode
capital to be at risk need a more realistic approach to project evalua­ risks as shown in the Table 8.
tion, risk management and project execution [52]. Thus, from the The overall criteria system of S-PHS project with above three
collected questionnaire data, the CRCs for DBFO mode are identified management modes can be presented in the Fig. 2.
according to their impact values as shown in Table 5.
Similarly, the risk criteria system of DBFO-S-PHS project can be
determined as the Table 6.

Table 5
CRCs selection of DBFO mode.
CRCs Pre. Pro. Sev. Con. Impact Nor. Ranking

Payment risk 7.327 8.015 8.612 7.045 7.726 1 1


Legal and policy risk 7.012 8.035 8.122 7.542 7.765 0.968 2
Government decision-making risk 6.988 7.568 8.114 7.452 7.520 0.893 3
Lack of supporting infrastructure 7.056 7.586 8.320 7.125 7.505 0.885 4
Concession period 7.324 6.772 8.683 7.125 7.443 0.852 5
No clear division of responsibilities and obligations 6.895 8.569 7.785 6.564 7.413 0.837 6
Imperfect supervision system 6.523 7.324 8.103 7.152 7.254 0.754 7
Lack of PPP experience 6.981 7.256 6.784 7.546 7.136 0.692 8

5
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Table 7
CRCs selection of BOO mode.
CRCs Pre. Pro. Sev. Con. Impact Nor. Ranking

Payment risk 7.056 8.782 8.125 7.656 7.879 1 1


Legal and policy risk 7.526 7.879 8.561 6.974 7.714 0.876 2
Lack of supporting infrastructure 7.871 7.568 8.421 6.890 7.667 0.841 3
Contract change risk 7.856 8.012 7.583 7.032 7.611 0.799 4
Government credit risk 6.746 7.387 8.154 8.046 7.562 0.762 5
No clear division of responsibilities and obligations 7.048 8.047 7.226 7.864 7.535 0.741 6
Government intervention 6.743 7.574 8.019 7.685 7.480 0.708 7
Lack of PPP experience 7.056 7.589 6.978 8.010 7.397 0.638 8

Table 8 where x in the universe U is called a normal cloud. The cloud model can
Risk criteria system of BOO-S-PHS project. effectively integrate the randomness and fuzziness of concepts and
First-class criteria Symbol Second-class criteria Symbol describe the overall quantitative property of a concept by three
numerical characteristics, namely, Expectation (Ex), Entropy (En),
Technical risk C1 Construction technology C11 and Hyper entropy (He).
Maintenance risk C12
Corrosion risk C13 Definition 3. [61]. Assume that there are two clouds X (E x1, En1, He1)
Economic risk C2 Payment risk C21 and Y (E x2, En2, He2), some operations between cloud X and Y can be
Revenue risk C22
defined as follows:
Inflation C23
Change in tax C24
X + Y = (E x 1 + E x 2, En1 2 + En2 2 , He1 2 + He2 2 ) (4)
Legal and policy risk C25
Contract change risk C26
Government C3 Government credit risk C31 X Y = (E x 1 E x 2, En1 2 + En2 2 , He1 2 + He2 2 ) (5)
Government intervention C32
Environment risk C4 Unforeseen weather or geotechnical C41
conditions X × Y = (E x 1 E x 2 , (En1 E x2) 2 + (E x1 En2 )2 , (He1 E x2) 2 + (E x1 He2 )2 )
Environmental pollution nearby risk C42 (6)
Social risk C5 Public opposition risk C51
Lack of supporting infrastructure C52 2 2
No clear division of responsibilities and C53 E x1 E x 1 En1 En2 E x1
X÷Y = , × + ,
obligations E x2 E x 2 E x1 E x2 E x2
Lack of PPP experience C54
Force majeure C55 2 2
He1 He2
× +
E x1 E x2
(7)
3. Basic theory and assessment framework
X m = (E x1, m E x1 m 1En1, m E x1 m 1He1) (8)
3.1. Basic theory
mX = (mE x1, m En1, m He1) (9)
The linguistic hesitant fuzzy sets can help DMs to give their eva­
luation in natural language in the fuzzy and complex environment to
make up the limitation of human cognition. Set S = {s0, ...,st 1} as a Actually, decision-makers are used to natural language instead of
linguistic term set, and t is an odd number. A LHFS is a set of S ap­ numerical values to assess problems [38].
pending the hesitation level valuing in[0, 1] as Definition 4. [62]. Supposing S = {s0, s1, ...,st 1} is the linguistic term
LH = {(s (i), lh (s (i) )|s (i) S )}, where lh (s (i ) ) = {r1, r2, ...,rmi} is a set set, t is an odd number, a valid universe [Xmin, Xmax] is given by experts.
with mi valuing in [0, 1] representing the possible membership degrees Correspondingly, the cloud Ai(Exi, Eni, Hei) is utilized to denotes the
of the element sθ(i) ∈ S to the set LH [59]. linguistic value si.
For example, Let linguistic term set be s0: very poor, s1: poor, s2:
slightly poor, s3: fair, s4: slightly good, s5: good, s6: very good. In the eva­ A golden method for generating five clouds was proposed by Wang
luation process, the expert commission may give a value of 0.6 for and Feng [63] based on 3σ principle of the normal distribution curve.
“low” and a value of 0.3 for “medium”, but hesitate to give a value of Zhu et al. [38] extended the five clouds into seven clouds by applying
0.2 or 0.3 for “very low”. In this situation, the evaluated value can be the golden ratio method as follows:
represented as LH = {(s1, 0.2, 0.3), (s2, 0.6), (s3, 0.3)} . Let linguistic term set be S = {s0: very poor , s1: poor , s2: slightly poor ,
Definition 1. [60]. U is Quantitative domain of exact numerical s3: fair , s4: slightly good, s5: good , s6: very good , } , where the universe is
representation, C is the qualitative concept of U. If x ∈ U and x is a [Xmin, Xmax], then seven clouds can be formulated by the golden radio
random implementation of C, and the certainty is μ(x) ∈ [0, 1], thus μ: method.
U → [0, 1], ∀x ∈ U, x → μ(x) is called cloud. Ex3 = (Xmin + Xmax )/2, Ex 0 = Xmin , Ex 6 = Xmax ,
Definition 2. Let U be the universe of discourse and T be a qualitative
Ex2 = Ex3 0.382(Xmax Xmin )/4, Ex 4 = Ex3 + 0.382(Xmax Xmin )/4,
concept in U. If x(x ∈ U) is a random instantiation of concept T, which
satisfies En′ ∼ N(En, He2), x ∼ N(Ex, En′2) and the certainty degree of x Ex1 = Ex3 (Xmax Xmin )/4, Ex5 = Ex3 + (Xmax Xmin )/4;
belonging to concept T satisfies Eq. (3).
En2 = En4 = 0.382(Xmax Xmin )/12, En3 = 0.618En4 ,
(x Ex )2
µ=e 2(En )2 (3) En1 = En5 = En4 /0.618, En 0 = En6 = En5/0.618;

6
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Fig. 2. Risk criteria system of S-PHS project with three typical management modes.

He2 = He4 = He3/0.618, He1 = He5 = He4 /0.618, He0 = He6 Definition 6. [38]. Let S = {s0, ...,st 1} be a linguistic term set and LH1
and LH2 be two LHFSs. The valid universe is [Xmin, Xmax]. The cloud Ai
= He5/0.618
(Exi, Eni, Hei) represents si. Then, the distance measure between LH1 and
where He3 is given by experts. LH2 is
Definition 5. [64]. Let S = {s0, s1, ...,st 1} be a linguistic term set. The
d (LH1, LH2)
valid universe is [Xmin, Xmax]. The cloud Ai(Exi, Eni, Hei) represents si.
* 1)2 + (HeLH
(EnLH * 1)2
Let LH = {s (i), lh (s (i) )|s (i) S } be a LHFS, where 1 * 1
ExLH
* 1)2 + (HeLH
(EnLH * 1)2 + (EnLH
* 2)2 + (HeLH
* 2 )2
lh (s (i ) ) = {r1, r2, ...,rmi} . * (ExLH
ALH * , ELH
* , HeLH
* ) is called the =
1
Xmax + Xmin
comprehensive cloud of LH. 1
* 2)2 + (HeLH
(EnLH * 2 )2
* 2
ExLH
* 1)2 + (HeLH
(EnLH * 1)2 + (EnLH
* 2)2 + (HeLH
* 2 )2
r r
(13)
relh (s (i) ) relh (s (i) )
* =
ExLH / Ex (i )
(i) index (LH ) lh (s (i ) )
(i ) index (LH ) lh (s (i ) )

(10)
And when EnLH
* 1 = HeLH
* 1 = EnLH
* 2 = HeLH
*2=0
r lh (s (i)) r r lh (s (i )) r
/ Ex (i ) En (i ) 1
(i ) index (LH ) lh (s (i ) ) (i ) index (LH ) lh (s (i ) ) d (LH1, LH2) = *1
ExLH *2
ExLH
* =
EnLH Xmax + Xmin (14)
r lh (s (i )) r r lh (s (i )) r
/ Ex
Proposition. Let LH1 and LH2 be two LHFSs. The following prop­
(i) index (LH ) lh (s (i) ) (i) index (LH ) lh (s (i ) ) ( i)

(11) erties hold:

* =
HeLH (He (i ) )
2 (i) 0 ≤ d(LH1, LH2) ≤ 1
(i) index (LH ) (12)
(ii) d (LH1, LH1) = 0
where |lh(sθ(i))| represents the total membership numbers to sθ(i) of lh (iii) d (LH1, LH2) = d (LH2, LH1)
(sθ(i)). And index (LH ) = { (i)|(s (i ), lh (s (i ) )) LH , LH (s (i) ) {0}} is
tenable in the situation of sθ(i) ∈ S. For aggregation of multiple clouds, a comprehensive cloud power

7
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

geometric averaging (CCPGA) operator is utilized. Finally, the shape and distance based cloud model (PDCM) simi­
Definition 7. [64]. Let Ω be the set of A i (Exi , Eni , Hei ), the CCPGA larity (Simc) can be determined as:
operator satisfies the mapping of CCPGA: Ωn → Ω: Simc = Simd × Sims (22)
n
CCPGA (A 1 , A 2 , ...,A n ) = (A i ) i
i=1

2
3.2. S-PHS risk assessment framework
n n n Eni
(Ex i ) i , (Ex i ) i
i × ,
i=1 i=1 i=1 Ex i In this section, the risk assessment framework of S-PHS project is
= established. For a comprehensive study, three typical PPP modes as
2
n n Hei BOT, DBFO as well as BOO are introduced as three scenarios to analyze
(Ex i ) i
i ×
i=1 i=1 Ex i the risk reduction of different management modes to the implementa­
= (Ex , En, He ) (15) tion of S-PHS project. Detail evaluation phrases are as follows.

=
1 + T (A i ) 3.2.1. Phrase I: Select the CRCs for the risk assessment of S-PHS project
i n
i=1
(1 + T (A i )) (16) The CRCs of S-PHS project risk assessment under typical modes are
n
selected in the Section 2 to separately distinguish the risks adhering to
T (A i ) = Sup (A i , A j ) S-PHS project itself and the modes we select. Besides, the selected
j =1 management mode would have certain impacts on the risk values of S-
(17)
PHS project in some degree. Namely, the same risk criteria under dif­
j i

Sup (A i , A j ) = 1 d (A i , A j ) (18) ferent management modes may gain assessment values with relatively
large gaps, which is based on the development level and expectation of
the combination of industry with management.
The Sup(Aαi, Aαj) is considered to be the support for Aαi from Aαj,
which satisfies the following properties: 3.2.2. Phrase II: Collect basic data of project risk
To deal with the complicated and uncertain S-PHS project risk as­
(i) Sup(Aαi, Aαj) ∈ [0, 1]; sessment issue and consider the hesitation of DMs and practicability of
(ii) Sup (A i , A j ) = Sup (A j , A i ) ; evaluation methods, LHFSs is adopted to describe risk information. In
(iii) Sup(Aαi, Aαj) ≥ Sup(Aαk, Aαt), if d(Aαi, Aαj) < d(Aαk, Aαt), where d is this phrase, an expert committee introduced in the Section 2 is invited
a distance measure between two LHFSs. to obtain the required basic data in the form of LHFSs under 4-di­
mention for realizing the objectives of this study. Then, consolidate all
If Sup (A i , A j ) = k for all i ≠ j, k is a real number, then the CCPGA information and formulate linguistic hesitation fuzzy decision matrix
will degenerate into comprehensive cloud geometric averaging (CCGA). M k = (mijk )p × q (k = 1, 2, 3) .
n 1
CCGA (A 1 , A 2 , ...,A n ) =
i=1
(A i ) n (19) 3.2.3. Phrase III: Comprehensive risk assessment of S-PHS project

Definition 8. Set two clouds as Ai(Exi, Eni, Hei) and Aj(Exj, Enj, Hej). The Step 1. Construct standard clouds.
similarly degree in shape (Sims) of these two clouds can be calculated
by Eq. (20). Seven standard clouds are calculated in the universe [0, 10] based
min(D (Ai ), D (Aj )) on 3σ principle of the normal distribution curve for risk assessment of S-
Sims (Ai , Aj ) = PHS in China under four scenarios. Due to the uncertainty and dis­
max(D (Ai ), D (Aj )) (20)
creteness of risk values, the standard clouds are classified into seven
where D (A) = + En2 He 2 parts as “extremely high (EH)”, “high (H)”, “middle high (MH)”,
Definition 9. Set two clouds as Ai(Exi, Eni, Hei) and Aj(Exj, Enj, Hej). d is “normal (N)”, “middle low (ML)”, “low (L)” as well as “extremely low
distance of the Ex between two clouds, and d = |Ex i Ex j |, and the (EL)” which can help describe the characteristics of risk more com­
distance based cloud model can be calculated as Eq. (21). prehensively.
S (d )
Simd = Step 2. Initial data processing
S (0) (21)
where S(d) is the overlapping area of expected curve of two clouds. S(0) Table 9
is the overlapping area when d = 0 . Fitting data.
In general, there is a certain corresponding relationship between
θ0 k a b c Adj R sq
Simd and the relative distance of the expectations of two clouds θ when
the ratio of entropy between two clouds k is determined. The re­ 0.0 1.000 1.42 -0.3142 0.527 1.0000
lationship satisfies the function y = a × exp( ((x b)/ c ) 2) which has 0.1 1.222 1.256 -0.2038 0.4772 0.9995
been proved by Jun et al. [65] and the fitting data can be shown in the 0.2 1.500 1.088 -0.08523 0.4213 0.9990
Table 9. 0.3 1.857 1.026 -0.01631 0.393 0.9992
0.4 2.333 1.007 0.02322 0.3859 0.9997
where Adj R sq is the Adjusted R-square. 0.5 3.000 1.002 0.048 0.3914 0.9999
And = |Ex i Ex j |/(3 × (Eni + Enj )), 0 = (k 1)/(k + 1). 0.6 4.000 1.001 0.06496 0.4043 0.9998
0.7 5.667 1.002 0.07731 0.4218 0.9996
Condition 1: When θ ∈ [0, 1), the Simd > 0; 0.8 9.000 1.002 0.08654 0.4423 0.9994
0.9 19.000 1.003 0.09325 0.4654 0.9992
Condition 2: When [1, + ), the Simd = 0 ;
1 20000 1.003 0.09759 0.4911 0.9993

8
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

To ensure the integrity of evaluation information and avoid ex­ in the step 4.
tending the fuzziness in the distance calculation process, LHFSs are
converted into cloud model based on Definition 5 through Step 4. Compute the support degree based on distance value.
Eqs. (10)–(12) for further handling.
Then the support degree (Sup) from Aαi to Aαj is calculated by
Step 3. Determine the distance of every two criteria. Eq. (18). α is the number of alternatives, i and j are two different cri­
teria. Then, the comprehensive support degree of each criteria can be
The distances of every two criteria are calculated by Eq. (16). And calculated from Sup in the step 6.
these distance values are the basis for calculating the degree of support

Fig. 3. The framework of S-PHS project risk assessment.

9
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Step 5. Determine the comprehensive support of each criteria. After handling risk criteria values into comprehensive cloud, the
distance between every two criteria should be determined by Eq. (13)
Calculate the comprehensive support T(Aαi) from Aαi to other cri­ which is utilized to calculate the Sup of each criteria by Eq. (18) and the
teria by Eq. (17). And the operator βi can be determined through comprehensive support degree T(Aαi) by Eq. (17) as well as the operator
Eq. (16) based on the comprehensive support value of each criteria. βi obtained by Eq. (16) as shown in the Table 11. Together with the
comprehensive support degree T(Aαi) and the operator βi, the overall
Step 6. Aggregate information and obtain the final risk values under risk degree of S-PHS project in China is calculated as (6.701, 0.460,
various modes. 0.405) by CCPGA operator, which has been represented in the standard
common set as the Fig. 5.
Aggregate above information into a comprehensive assessment re­ From the calculation result we can see that the risk level of S-PHS
sult by CCPGA operator. Analyze risk value of S-PHS project and obtain project nearly locates between “MH” and “H”. Thus, overall response
the risk level of each criteria in detail. strategies for all criteria are necessary in order to deal with risks rea­
sonably.
Step 7. Compute Simc.
4.1. Scenario I: Risk assessment of S-PHS project under the BOT
Compute the Simc between evaluation object and each evaluation management mode
standard to accurately locate risk level. When the obtained k and θ0
values are in the medium value of the smallest partition area in the Adopting BOT management mode in S-PHS project indicates that
table, we use the maximum approximation method and the mean the risk criteria system of BOT-S-PHS project will be different from it
method to determine the parameters of the fitting curve. under single S-PHS project. Moreover, some criteria values of S-PHS
project will be influenced by the mode type we take to a certain extent.
Step 8. Put forward risk response of all criteria. After calculation through the proposed risk assessment framework
in this paper, a comprehensive final risk degree is calculated as (6.175,
Based on the risk level of S-PHS project, risk degree of each risk 0.404, 0.359) and shown together with the standard cloud in the Fig. 6.
criteria as well as the comprehensive similarity value of Simc of the Besides, T(Aαi), βi and cloud values are shown in the Appendix B and C,
project, the corresponding risk allocation and risk response are for­ respectively.
mulated through literature review, book study and expert consultation It can be clearly seen that the risk degree of BOT-S-PHS project
for further promoting the smooth implementation of S-PHS project in locates in the middle of “MH” and “H” with high probability. But when
China. compared to single S-PHS model, the risk value decreases in the value
The overall risk assessment flow of this paper is shown in Fig. 3. of Ex. That is to say, BOT mode can help reduce the risk value of S-PHS
project to a certain degree. Nevertheless, risk response measures are
4. Risk assessment of S-PHS in various scenarios in China also necessary when it comes to build a S-PHS project under the BOT
mode with limited resources since such project still in high risk level.
Along with more and more attention has been paid on the risks of
construction project, this section will assess the risk degree of PPP-S- 4.2. Scenario II: Risk assessment of S-PHS project under the DBFO
PHS project through the proposed model to provide a reference on the management mode
project construction and operation and promote its successful and
sustainable development. First, the expert decision group is required to DBFO is a typical and widely used management mode in the
obtain the fuzzy basic risk data based on their work experience and worldwide. In this part, the effectiveness of adopting DBFO manage­
academic background as well as the current development situation of ment mode on S-PHS project implementation is calculated and ana­
China via a discussion meeting. In the first part of the meeting, the lyzed. Risk assessment criteria system are shown in the Table 6. The
definition of each potential criteria is given as a reference in case ex­ original evaluation data, T(Aαi), βi and cloud values in the evaluation
perts in different fields cannot fully understand the indicators. process are shown in the Appendix A–C, respectively. Similar to the S-
Secondly, the expert decision group is requested to give a comprehen­ PHS risk assessment process, the final risk degree of DBFO-S-PHS pro­
sive risk evaluation language towards each factor after discussion. ject is (6.281, 0.411, 0.363) which is mainly between “MH” and “H” as
Thirdly, according to the calculated final risk value, the risk response
strategies are discussed and determined. The original risk criteria va­
lues in 4-demention for S-PHS project are gained and shown in the
Appendix A.
Then, the standard comment sets referring to “EH”, “H”, “MH”, “N”,
“ML”, “L” and “EL” as S = {s0, s1, s2, s3, s4, s5, s6} are established for the
index system through the golden radio method as shown in the Fig. 4
with a universe of [0, 10], and He3 = 0.1 given by expert committee.
Generally, seven risk languages can fully classify and describe risk le­
vels comprehensively which is very convenient for experts to evaluate
risks more practically.
In order to ensure the integrity of the risk information and the ac­
curacy of calculation, original LHFSs are converted into comprehensive
cloud for further comprehensive aggregation by Definition 5. Further­
more, the 4-dimension cloud models can be aggregated into one cloud
through CCGA operator as Eq. (22). And the basic data of cloud models
are presented in the Table 10. Fig. 4. Comment set.

10
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Table 10
S-PHS project cloud value of each criteria.
Criteria RPM RPO RCS RUC Aggregated cloud

C11 (5.94, 0.38, 0.35) (6.00, 0.34, 0.32) (6.88, 0.45, 0.31) (6.84, 0.44, 0.31) (6.40, 0.40, 0.33)
C12 (5.70, 0.37, 0.35) (7.93, 0.61, 0.52) (7.11, 0.57, 0.52) (5.29, 0.36, 0.46) (6.42, 0.47, 0.47)
C13 (6.00, 0.34, 0.32) (7.91, 0.60, 0.52) (7.27, 0.52, 0.52) (5.63, 0.44, 0.46) (6.64, 0.47, 0.46)
C21 (7.02, 0.46, 0.31) (7.66, 0.57, 0.52) (7.88, 0.60, 0.52) (6.70, 0.43, 0.32) (7.30, 0.51, 0.42)
C22 (7.53, 0.55, 0.52) (7.24, 0.53, 0.52) (6.01, 0.40, 0.35) (6.60, 0.42, 0.32) (6.82, 0.47, 0.43)
C23 (7.35, 0.52, 0.52) (6.44, 0.41, 0.28) (7.5, 0.52, 0.26) (7.30, 0.52, 0.52) (7.14, 0.49, 0.41)
C24 (6.01, 0.40, 0.35) (7.68, 0.58, 0.52) (4.86, 0.27, 0.25) (5.71, 0.31, 0.32) (5.98, 0.38, 0.35)
C31 (8.01, 0.61, 0.52) (6.14, 0.37, 0.32) (6.33, 0.43, 0.46) (7.83, 0.59, 0.52) (7.03, 0.49, 0.46)
C32 (6.04, 0.40, 0.35) (7.27, 0.52, 0.52) (6.18, 0.41, 0.35) (6.84, 0.58, 0.52) (6.56, 0.47, 0.44)
C41 (5.75, 0.39, 0.35) (5, 0.27, 0.25) (6.78, 0.43, 0.31) (6.97, 0.50, 0.51) (6.07, 0.39, 0.35)
C42 (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26)

Table 11
The T(Aαi) and βi of S-PHS project.
Criteria T(Aαi) βi

C11 9.019 0.0559


C12 16.117 0.0567
C13 15.404 0.0543
C21 15.919 0.0560
C22 15.723 0.0554
C23 15.774 0.0555
C24 16.090 0.0566
C31 15.349 0.0541
C32 15.511 0.0547
C41 15.258 0.0538
C42 16.598 0.0550

Fig. 7. Risk degree of DBFO-S-PHS project.

Fig. 8. Risk degree of BOO-S-PHS project.


Fig. 5. S-PHS project risk degree.
shown in the Fig. 7. Comparing to single S-PHS project, there is an
obvious downward trend in risk value. But when compared with S-PHS
project under the BOT mode, the risk of DBFO-S-PHS project is rela­
tively high. In a sense, BOT is superior to DBFO mode in risk reduction
of S-PHS projects. Furthermore, when it comes to construct a DBFO-S-
PHS project, effective risk response strategies should be proposed so as
to promote its smooth progress.

4.3. Scenario III: Risk assessment of S-PHS project under the BOO
management mode

In this scenario, we adopt BOO management mode into S-PHS


project. Distinctive risk criteria system is established in the Table 8.
Through the proposed evaluation framework, the risk degree of BOO-S-
PHS project can be determined as shown in the Fig. 7. The detail data
information in the evaluation process is shown in the Appendix A–C. It
can be clearly demonstrated that BOO-S-PHS project risk is (6.271,
0.419, 0.372) which nearly locates in the middle of “MH” and “H” as
Fig. 6. Risk degree of BOT-S-PHS project. shown in the Fig. 8. By comparison, project risk under BOO mode is

11
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Fig. 9. The risk degree of S-PHS project under typical management modes.

Table 12 According to the principle of maximum proximity, the four sce­


Similarity degree of evaluation to each standard. narios of S-PSH project is more closed to S4 with the similarities of
Scenario S0 S1 S2 S3 S4 S5 S6
0.429, 0.824, 0.708 and 0.741 as shown in the Table 12, which means
the scenario of BOT-S-PHS is more benefit for risk reduction and S-PHS
S-PHS 0 0 0 0.009 0.429 0.435 0.008 project construction. More specifically, scenario BOT-S-PHS has the
BOT-S-PHS 0 0 0.002 0.063 0.834 0.161 0 highest similarity degree to “MH”, a similarity of 0.063 to “N” and
DBFO-S-PHS 0 0 0 0.037 0.708 0.212 0.002
BOO-S-PHS 0 0 0 0.052 0.741 0.205 0.158
0.002 to “ML”. Besides, scenario BOO-S-PHS has the second highest
similarity degree to “MH”, a similarity degree of 0.052 to “N” and 0 to
“ML”. The scenario DBFO-S-PHS has the third similarity degree to
higher than the one under the BOT mode but lower than it under the “MH”, 0.037 to “N” and 0 to “ML”. But for single S-PHS, the similarity
DBFO mode. Consequently, the BOO mode is more beneficial to S-PHS to “MH” is slightly higher than the one to “H”. That is to say, the risk
project than DBFO mode in the aspect in risk reduction which can be level of S-PHS projects in China approximately gets the median value of
considered as a second management mode choice for decision makers. “MH” and “H”.
But overall, the risk level of BOO-S-PHS project is still high, so it is
necessary to put forward response strategies to each risk factor.
4.5. Comparative analysis
4.4. Comprehensive analysis
To prove the accuracy and superiority of the proposed risk assess­
Summarize the above four situations, the overall risk evaluation ment framework, a comparative analysis is conducted in this section.
result can be shown in the Fig. 9. It can be clearly seen from the picture The risk calculation model proposed in this paper includes three key
that the risk level of all scenarios nearly locates in the middle of “MH” steps as risk representation method, fuzzy data conversion and risk
and “H”. But in the terms of the Ex value of each cloud model, there are element aggregation. Due to the ambiguity of risk data and the hesi­
obvious differences among scenarios. In order to describe the risk level tation of decision makers, the data representation method as HTFLs
of each scenario more accurately, this paper calculates the similarity combining with cloud model has obvious advantages in theory and
degrees between the evaluation result and the standard through the practical application. Therefore, two types comparative analysis are
following steps for further analysis and the results are showed in the conducted in this section as data conversion comparison and risk ag­
Table 12. gregation comparison to prove the accuracy and superiority of the risk
assessment model established in this paper.
Step 1. Calculate the Sims of two clouds based on entropy and hyper- In the aspect data conversion from linguistic language to compu­
entropy. table data, Zhu et al. [38] proposed a different computational approach
Step 2. Compute the entropy ratio of the two cloud models k, and as follows.
then determine the value of θ0. Definition 10. [38]. Let S = {s0, s1, ...,st 1} be a linguistic term set. The
Step 3. According the calculation principle and the Table 9, obtain valid universe is [Xmin, Xmax]. The cloud Ai(Exi, Eni, Hei) represents si.
the Simd. Let LH = {s (i), lh (s (i) )|s (i) S } be a LHFS, where
Step 4. Formulate comprehensive similarity degree Simc according lh (s (i ) ) = {r1, r2, ...,rmi} . ALH (ExLH , ELH , HeLH ) is called the
to the Eq. (25). comprehensive cloud of LH.

12
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

He* = 0.2622 + 0.1622 + 0.12 = 0.324


1 Ex (i )
ExLH = r
index (LH ) lh (s (i ) ) By Definition 10, the comprehensive cloud model is A.
(i) index (LH ) r lh (s (i) )
(23)
1 Ex (i )
1 2 ExLH = r
EnLH = (En (i ) ) index (LH ) lh (s
index (LH ) (i) index (LH ) (i ) ) r lh (s (i) )
(i) index (LH ) (24)
3.09
1 = 1/3 (0.2 + 0.3) + 4.045*0.6 + 5*0.3 = 1.5665
HeLH = (He (i ) )
2 2
index (LH ) (i) index (LH ) (25)
1 2
EnLH = (En (i ) )
index (LH ) (i) index (LH )

By comparison, there are virtual differences between Definitions 5 1


= (0.3932 + 1.032 + 0.6372) = 0.7351
and 10 in Linguistic hesitant fuzzy sets conversion. For example, let 3
linguistic term set be S = {s0: very poor , s1: poor , s2: slightly poor ,
s3: fair , s4: slightly good, s5: good , s6: very good , } . The valid universe is 1 2
HeLH = (He (i ) )
[Xmin, Xmax]. In the evaluation process, the expert commission may give index (LH ) (i) index (LH )
a value of 0.6 for “low” and a value of 0.3 for “medium”, but hesitate to
1
give a value of 0.2 or 0.3 for “very low”. In this situation, the evaluated = (0.12 + 0.2622 + 0.1622) 0.187
value can be represented as LH = {(s1, 0.2, 0.3), (s2, 0.6), (s3, 0.3)} , then 3
converting the evaluation information into comprehensive cloud Compare the two cloud models in one map as the Fig. 10, we can see
model. that there are obvious differences. By Definition 10, the comprehensive
By Definition 5, the comprehensive cloud model is cloud deviates greatly from the center of the standard evaluation lan­
A* = (4.087, 0.624, 0.324) . The calculation process is as follows. guage. On the contrary, this phenomenon can be reasonable solved and
the evaluation results can be more truthfully reflected by Definition 5.
0.2 + 0.3 0.2 + 0.3
Ex* = / + 0.6 + 0.3 *3.09 Thus, the data conversion approach adopted in this paper process
2 2
stronger rationality and application value.
0.2 + 0.3 In the aspect of risk aggregation method, fuzzy comprehensive
+ 0.6/ + 0.6 + 0.3 *4.045
2 evaluation (FSE) is a mature MCDM method with widespread utiliza­
0.2 + 0.3 tion in risk assessment, investment decision making, site selection and
+ 0.3/ + 0.6 + 0.3 *5 = 4.087
2 so on. Thus, this paper uses FSE model to comparatively analyze the
risk level of S-PHS project in China. The evaluation result can be clearly
( 0.2 + 0.3
2 (
/
0.2 + 0.3
2
+ 0.6 + 0.3 ) )*3.09*1.03 shown as the Table 13.
From the calculation result we can see that no matter which eva­
( (
+ 0.6/
0.2 + 0.3
2
+ 0.6 + 0.3) )*4.045*0.637 luation model, the risk result of PPP-S-PHS project is between M and
MH, which indicates that the evaluation result is reliable. However, the
+ (0.3/ ( + 0.6 + 0.3) )*5*0.393
0.2 + 0.3
2 risk ranking of DBFO-S-PHS and BOO-S-PHS are exchanged when use
En* = = 0.624 the FSE model since DBFO-S-PHS process relatively more average cri­
( /( 0.2 + 0.3
2
0.2 + 0.3
2
+ 0.6 + 0.3) )*3.09 teria values but BOO-S-PHS is prominent in a certain type of criteria.
The reason for this phenomenon is that attributes are completely
+ (0.6/ ( + 0.6 + 0.3) )*4.045
0.2 + 0.3
2 compensated by each other in the FSE method, which can't reflect the
+ (0.3/ ( + 0.6 + 0.3) )*5 real function of each attribute. But this shortcoming can be effectively
0.2 + 0.3
2
avoided by the Cloud model utilized in this paper through calculating

Fig. 10. The comparative cloud between Definitions 5 and 10.

13
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Table 13
Risk result formulated by fuzzy comprehensive evaluation method.
S-PHS BOT-S-PHS DBFO-S-PHS BOO-S-PHS

FSE model (6.715,0.461,0.398) (6.120,0.406,0.357) (6.323,0.413,0.360) (6.298,0.420,0.369)


Ranking 4 1 3 2
Cloud model (6.701,0.460,0.405) (6.175,0.404,0.359) (6.281,0.411,0.363) (6.271,0.419,0.372)
Ranking 4 1 2 3

Table 14
Risk allocation scheme of PPP S-PHS project.
No Risk factors Risk allocation
Public Private The third party (Insurance company)

1 Construction technology √ √
2 Maintenance risk √ √
3 Corrosion risk √ √
4 Payment risk √ √
5 Revenue risk √ √
6 Inflation √ √
7 Change in tax √
8 Government decision-making risk √
9 Legal and policy risk √
10 Government intervention √
11 Government corruption √
12 Unclear division of responsibilities and obligations √ √ √
13 Concession period √ √
14 Imperfect supervision system √
15 Government credit risk √
16 Contract change risk √ √
17 Lack of PPP experience √ √
18 Unforeseen weather or geotechnical conditions √
19 Environmental pollution √ √ √
20 Public opposition √ √
21 Lack of supporting infrastructure √
22 Force majeure √

the distance between every two criteria and adding criteria support supervision system, government credit risk, unforeseen weather or
operator in the calculation process. In comparison, cloud model ag­ geotechnical conditions as well as lack of supporting infrastructure. For
gregation operator can more scientifically aggregate fuzzy sets with comprehensively ameliorate the risks of these factors, certain responses
great hesitation without losing information in the calculation process. are proposed as follows.
In conclusion, the risk assessment model in this paper is accurate and
more effective and applicable. (i) Change in tax.

5. Risk response According to the book of Public Private Partnerships Theory and
Practice, change in tax is the risk which should be undertaken by
Despite the increasing research on S-PHS worldwide, the risk level government section. Change in tax is normal in renewable energy de­
of S-PHS is still in the middle of “MH” and “H” even if multiple PPP velopment since the government has different incentive plans as dif­
modes are added for risk sharing. Hence, effective risk response is in­ ferent stages. But the change in tax can bring great fluctuation in
dispensable and necessary to ensure the success of this project. First, smooth operation of PPP renewable energy project. To deal with it,
through literature research, related book studies and expert meetings, public could broaden and encourage effective and suitable financial
the risk allocation scheme for Chinese S-PHS project under above three instruments, and improve national core competitiveness through con­
PPP modes has been determined, as shown in Table 14 tinuously investing in technological innovation and improvement.
[48,52,58,66,67]. Besides, government may better change the tax based on the domestic
As literature review, risk response measures should be proposed in project affordability from a global perspective.
accordance with corresponding risk allocated parties and the evaluation
results in order to provide a reference for the public sectors, private (ii) Government decision-making risk.
sectors, the third party (insurance company) and sharing sections. The
corresponding risk response can be shown as follows. In preparation, government should conduct enough discussion as
well as communication with private sectors on the key procedures or
➢ Public unilateral risk response times of S-PHS project to formulate an adequate plan and avoid the
decision risks arising from non-communication. In the medium term,
The public sector is mainly directed at government behavior, in­ the combination of foreign successful experience with the reality of our
cluding change in tax, government decision-making risk, legal and country is necessary in government decision process to ensure pro­
policy risk, government intervention, government corruption, imperfect gressiveness and rationality of decision results.

14
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

(iii) Legal and policy risk. the amount of payment.


(ii) Revenue risk. Revenue risk always comes from every part of PPP
Government should accurately analyze the orientation of PPP S-PHS project such length of franchise period as well as participants’
project in order that a set of comprehensive policy can be formulated fault, change in tax and electricity price, etc. Thus, revenue risk
from a holistic perspective so that temporary changes in polices can be can be effectively avoided through overall planning and careful
effectively avoided, which has great adverse effects on the normal de­ study before the project starts. A certain of experts can be invited
velopment of PPP S-PHS project. to determine the most suitable franchise period length and the
price of electricity. Furthermore, effective information commu­
(iv) Government intervention/ Government corruption/ Government nication system should be applied to ensure the timely disposal of
credit risk. the emergences to minimize the revenue losses.
(iii) Inflation. Inflation can be influenced by numerous factors and the
Establish a third-party organization to supervise the government alteration of it can cause cost or price changes which will have
and private sectors to ensure there is either no intervention or cor­ great impact on the smooth operation of PPP S-PHS project.
ruption from government in the period of franchise by which govern­ Fortunately, advance price coefficient setting plan can be effective
ment can strongly increase individual credit degree. The supervision when inflation occurs. Besides, some government regulation and
system can be made up by some prestigious professors or several guidance measures are necessary to improve the rational level of
masses. A written guarantee is necessary too. capital flow.
(iv) Concession period.
(v) Unforeseen weather or geotechnical conditions.
There is greater uncertainty in the concession period which may
This risk can be decreased in a certain degree at the stage of site greatly influence the benefits of the private section and public section.
selection and prospecting. In the process of construction and operation, Thus, the concession period should be reasonable planned and clearly
some prevention measures such as effective timely seismic monitoring stated in the contract. Besides, if the concession period has to be
or installation of natural disaster warning device as well as the early changed, the corresponding punishment and compensation mechanism
warning treatment should be formulated which is responsible by the and the subsequent operation mechanism must be clearly determined.
government.
(v) Contract change risk
(vi) Lack of supporting infrastructure.
Contract change covers a wide range of contents in general. Namely,
For PPP project, local government is responsible for ensure the a little bit change in the contract may lead to great impact. Hence,
electricity and land acquisition, and it is necessary for government to proper risk response will help reduce the uncertain risk of one project
communicate with relevant departments for project smooth running. If and implement it as planned. First, all parties should consider all kinds
the government fails to fulfill its corresponding duties, after supervision of unexpected situations as comprehensively as possible in advance,
and verification, the concession period should be extended. and clearly study their coping strategies. If the contract has to be
amended during the performance of one contract, multi-party con­
➢ The third party (insurance company) unilateral risk response sultation and discussion are needed to minimize the overall loss.

The main responsibility provided by the third-party containing in­ (vi) Lack of PPP experience
surance company is supervision and management in the construction
and operation process and collaborative risk response after risks oc­ Firstly, fair tendering in real sense should be achieved in selecting
currence. private sector. Moreover, whether the private sector has done at least
one successful PPP project should be an important assessment criterion
(i) Force majeure of public sector, which can give private sector certain lessons.

Build early warning system and timely information transmission (vii) Public opposition.
system so as to ensure the force majeure can be greatly prevented in the
prevention stage and reduce its occurrence possibility, increase the A stable social order is the prerequisite for the development of a
means to control if it occurs as well as reduce the impact. Besides, the project. Thus, appease the local people with certain preferential policies
third party should take the responsibilities of the supervision of system on electricity consumption, which will effectively solve the problem of
operation public opposition.

➢ Public and private joint responsibility response ➢ Private and the third party (insurance company) joint responsibility
response
Risks like payment risk, revenue risk, inflation, concession period,
contact change risk, lack of PPP experience and public opposition are The risks allotted to private and the third party (insurance com­
within the scope of joint responsibility of the public and private who pany) are construction technology, maintenance risk and corrosion risk
should take suitable risk response measures to ensure the normal op­ according to literature reviews and relevant book study.
eration of PPP S-PHS project.
(i) Construction technology
(i) Payment risk. For minimizing the payment risk, a clearly conces­
sion agreement should be formulated with the detail payment As no practical experience on S-PHS project in China, learning from
process as payment time, payment place, payment mode as well as foreign successful experience and technology is indispensable. Besides,

15
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

expert research teams can be established by private sector to study and foreign successful S-PHS project and Chinese actual condition, the key
innovate advanced technology in each period of PPP S-PHS project risk factors of S-PHS project and three typical PPP management modes
operation. Besides, the third party (insurance company) can broaden as BOT, DBFO and BOO are identified. There are 11 risks factors in the
their service channel and strengthen the supervision of existing tech­ final S-PHS project risk assessment criteria system and 18 factors in the
nology application safety and the supervision of expert research team. one under each typical PPP management mode, which can be classified
into five groups as technical risk, economic risk, participant risk, gov­
(ii) Maintenance risk ernment risk and social risk. Following that, a risk assessment frame­
work as LHFSs based cloud model is established to comprehensively
Large scale maintenance can increase the cost. Thus, effective pre­ calculate the current risk level of S-PHS project in China.
vention and timely detection of the problem can decrease the prob­ The result shows that S-PHS risk level in China locates in the middle
ability of risk occurrence. The third party mainly provides supervisory of “MH” and “H”, and BOT-S-PHS gains the lowest risk value comparing
functions and urges private sector to correct existing deficiencies. with other types. Furthermore, the superiority of PPP mode for S-PHS
project can be clearly demonstrated by the similarity degree compar­
(iii) Corrosion risk ison. On this basis, the risk response strategies for all risk factors in each
mode are put forward. Compared with existing research, the proposed
As seawater has a certain corrosive effect on environment or ma­ model in this paper has the following advantages: (1) The risk degree of
chine, research on anticorrosion technology is indispensable. In addi­ S-PHS project is firstly assessed by the proposed model in this paper; (2)
tion, the areas where plants or farms are numerously grown should be A 4-dimension risk value determination pattern is analyzed in detail;
avoided in the stage of site selection. Moreover, some suitable and (3) Critical risk factors of S-PHS project and three typical PPP man­
economic methods which can decrease seawater corrosion degree can agement modes as BOT, DBFO and BOO are identified; (4) A risk as­
be rationally utilized. And the third party should strengthen the su­ sessment framework of LHFSs based cloud model is utilized which are
pervision of the integrity of all process to effectively prevent the oc­ accessible to the facts with the characteristics as fuzziness and ran­
currence of risks. domness; (5) The superiority of PPP mode in risk reduction to S-PHS
project has been verified through scenarios analysis and similarity de­
➢ Public, Private and the third party (insurance company) trilateral gree comparison. Besides, DMs can well understand and endorse the
joint responsibility response. process decision making and the evaluation result. Overall, the pro­
posed risk assessment framework can extend the study fields of PPP S-
After risk sharing, unclear division of responsibilities and obliga­ PHS project and overcome the deficiencies in previous research.
tions, environmental pollution as well as force majeure are the risks Although there are a lot of advantages in this paper, certain im­
factors which should be handled by three parties as public, private and provements are still existed. Due to there are many management modes
the third party (insurance company) sections. besides PPP, such as EPC (Engineering Procurement Construction), it is
meaningful to further assess the risk level in other modes to search for
(i) Unclear division of responsibilities and obligations discrimination of the advantages of each mode in the application of S-
PHS project. If we can solve these problems in the future, the decision
At the beginning of the project, a clear division contract of re­ result would be more effective.
sponsibilities and obligations should be defined and approved by each
participant. Everyone had the right to supervise all participants’ beha­
CRediT authorship contribution statement
vior. Some fine or legal responsibilities can be pursued if any party
breach the contract stipulations.
Yunna Wu: Conceptualization, Resources, Supervision. Ting
Zhang: Methodology, Investigation, Writing - original draft. Kaifeng
(ii) Environmental pollution
Chen: Writing - review & editing. Liqi Yi: Validation, Visualization.
Environmental pollution is the main reason for stopping the service
of the first S-PHS project in Japan. Thus, in the process of site selection, Declaration of Competing Interest
public and private sector should try to avoid building upper reservoirs
where vegetation is abundant. Besides, environmental compensation
We declare that we have no financial and personal relationships
measures such as building protection forests outside a certain area of
with other people or organizations that can inappropriately influence
the upper reservoir or establishing appropriate isolation zones around
our work, there is no professional or other personal interest of any
the upper reservoir to prevent sea water splash. In the course of op­
nature or kind in any product, service and/or company that could be
eration, the third party should act as a strict supervisor to prevent
construed as influencing the position presented in, or the review of, the
private or public sectors from failing to operate in accordance with
manuscript entitled.
regulations.

6. Conclusions Acknowledgments

This paper analyzes the risk degree of S-PHS project and the one This research is supported by the National Social Science Fund of
under three typical management modes through a scientific and com­ China (19AGL027), the Fundamental Research Funds for the Central
prehensive framework as LHFSs based cloud model. Firstly, on the basis Universities (No. 2020MS066), the 2017 Special Project of Cultivation
of literature review, expert discussion meeting, the experience of and Development of Innovation Base (NO. Z171100002217024).

16
Appendix A
Y. Wu, et al.

Table 15
Criteria risk values in 4-dimention for S-PHS project.
RPM RPO RCS RUC

C11 LH = {(s2, 0.2, 0.3), (s4, 0.6), (s5, 0.3)} LH = {(s3, 0.3, 0.4), (s4, 0.5), (s5, 0.2, 0.3)} LH = {(s4, 0.4), (s5, 0.6)} LH = {(s4, 0.4, 0.5), (s5, 0.6)}
C12 LH = {(s2, 0.3), (s4, 0.5, 0.6), (s5, 0.2)} LH = {(s4, 0.3), (s5, 0.5, 0.6), (s6, 0.4)} LH = {(s2, 0.2), (s5, 0.4, 0.5), (s6, 0.1, 0.2)} LH = {(s2, 0.2, 0.3), (s3, 0.5, 0.6), (s6, 0.1)}
C13 LH = {(s3, 0.3, 0.4), (s4, 0.6), (s5, 0.2, 0.3)} LH = {(s4, 0.3), (s5, 0.5, 0.7), (s6, 0.4)} LH = {(s4, 0.4), (s5, 0.55, 0.65), (s5, 0.1, 0.2)} LH = {(s2, 0.3, 0.4), (s3, 0.5), (s6, 0.2)}
C21 LH = {(s3, 0.2, 0.3), (s5, 0.5, 0.6)} LH = {(s4, 0.3), (s5, 0.4, 0.5), (s6, 0.2, 0.3)} LH = {(s4, 0.2, 0.3), (s5, 0.7), (s6, 0.3, 0.4)} LH = {(s3, 0.1, 0.2), (s4, 0.2, 0.3), (s5, 0.5, 0.6)}
C22 LH = {(s4, 0.3), (s5, 0.6, 0.7), (s6, 0.2)} LH = {(s6, 0.1, 0.3), (s4, 0.4, 0.5), (s5, 0.1)} LH = {(s2, 0.2, 0.3), (s5, 0.3, 0.4), (s4, 0.5)} LH = {(s3, 0.1, 0.2), (s4, 0.4), (s5, 0.5, 0.6)}
C23 LH = {(s4, 0.3, 0.4), (s5, 0.6), (s6, 0.1, 0.2)} LH = {(s3, 0.4), (s5, 0.5, 0.6)} LH = {(s5, 0.5, 0.7)} LH = {(s4, 0.4), (s5, 0.6, 0.7), (s6, 0.1, 0.2)}
C24 LH = {(s2, 0.2, 0.3), (s4, 0.5), (s5, 0.3, 0.4)} LH = {(s4, 0.3, 0.4), (s5, 0.5), (s6, 0.3)} LH = {(s2, 0.4), (s3, 0.4), (s4, 0.2, 0.3)} LH = {(s3, 0.4, 0.5), (s4, 0.2), (s5, 0.1, 0.2)}
C31 LH = {(s4, 0.2), (s5, 0.5, 0.6), (s6, 0.3, 0.4)} LH = {(s4, 0.2, 0.3), (s3, 0.5), (s5, 0.4, 0.5)} LH = {(s3, 0.3), (s4, 0.4), (s6, 0.1, 0.2)} LH = {(s5, 0.5), (s4, 0.2), (s6, 0.2, 0.3)}
C32 LH = {(s2, 0.2), (s4, 0.4, 0.5), (s5, 0.3)} LH = {(s4, 0.4), (s5, 0.5), (s6, 0.1, 0.2)} LH = {(s2, 0.2), (s4, 0.4, 0.5), (s5, 0.4)} LH = {(s2, 0.4), (s5, 0.5), (s6, 0.2, 0.3)}
C41 LH = {(s4, 0.3, 0.4), (s2, 0.3), (s5, 0.2, 0.3)} LH = {(s2, 0.3), (s3, 0.4), (s4, 0.3)} LH = {(s4, 0.3, 0.4), (s5, 0.4)} LH = {(s3, 0.3, 0.4), (s5, 0.4, 0.5), (s6, 0.1, 0.2)}
C42 LH = {(s5, 0.5)} LH = {(s5, 0.5)} LH = {(s5, 0.5)} LH = {(s5, 0.5)}

17
Table 16
Criteria risk values in 4-dimention for S-PHS project under the BOT mode.
RPM RPO RCS RUC

C11 LH = {(s2, 0.2, 0.3), (s4, 0.6), (s5, 0.3)} LH = {(s3, 0.3, 0.4), (s4, 0.5), (s5, 0.2, 0.3)} LH = {(s4, 0.4), (s5, 0.6)} LH = {(s4, 0.4, 0.5), (s5, 0.6)}
C12 LH = {(s2, 0.3), (s4, 0.5, 0.6), (s5, 0.2)} LH = {(s4, 0.3), (s5, 0.5, 0.6), (s6, 0.4)} LH = {(s1, 0.2), (s3, 0.4), (s4, 0.3, 0.4)} LH = {(s2, 0.5), (s3, 0.2, 0.3), (s4, 0.1)}
C13 LH = {(s3, 0.3, 0.4), (s4, 0.6), (s5, 0.2, 0.3)} LH = {(s4, 0.3), (s5, 0.5, 0.7), (s6, 0.4)} LH = {(s4, 0.4), (s5, 0.55, 0.65), (s5, 0.1, 0.2)} LH = {(s2, 0.3, 0.4), (s3, 0.5), (s6, 0.2)}
C21 LH = {(s3, 0.2, 0.4), (s5, 0.5)} LH = {(s2, 0.1), (s4, 0.2, 0.3), (s5, 0.65)} LH = {(s3, 0.3), (s4, 0.5), (s5, 0.3, 0.4)} LH = {(s2, 0.3), (s4, 0.4, 0.5), (s5, 0.1, 0.2)}
C22 LH = {(s4, 0.5), (s5, 0.4, 0.5)} LH = {(s2, 0.3), (s4, 0.4, 0.5), (s5, 0.1)} LH = {(s2, 0.2, 0.3), (s3, 0.2), (s4, 0.5)} LH = {(s3, 0.2), (s4, 0.4), (s5, 0.5)}
C23 LH = {(s4, 0.3, 0.4), (s5, 0.6), (s6, 0.1, 0.2)} LH = {(s3, 0.4), (s5, 0.5, 0.6)} LH = {(s5, 0.5, 0.7)} LH = {(s4, 0.4), (s5, 0.6, 0.7), (s6, 0.1, 0.2)}
C24 LH = {(s2, 0.2, 0.3), (s4, 0.5), (s5, 0.3, 0.4)} LH = {(s4, 0.3, 0.4), (s5, 0.5), (s6, 0.3)} LH = {(s2, 0.4), (s3, 0.4), (s4, 0.2, 0.3)} LH = {(s3, 0.4, 0.5), (s4, 0.2), (s5, 0.1, 0.2)}
C25 LH = {(s4, 0.3, 0.4), (s5, 0.5), (s6, 0.2, 0.3)} LH = {(s3, 0.3), (s4, 0.5), (s5, 0.3, 0.4)} LH = {(s3, 0.2, 0.3), (s4, 0.7)} LH = {(s2, 0.3, 0.4), (s3, 0.5), (s4, 0.4)}
C31 LH = {(s4, 0.2), (s5, 0.5, 0.6), (s6, 0.3, 0.4)} LH = {(s2, 0.2, 0.3), (s3, 0.6), (s4, 0.1, 0.2)} LH = {(s3, 0.3), (s4, 0.4), (s6, 0.1, 0.2)} LH = {(s5, 0.5), (s4, 0.2), (s6, 0.2, 0.3)}
C32 LH = {(s2, 0.2), (s4, 0.4, 0.5), (s5, 0.3)} LH = {(s4, 0.4), (s5, 0.5), (s6, 0.1, 0.2)} LH = {(s2, 0.2), (s4, 0.4, 0.5), (s5, 0.4)} LH = {(s2, 0.4), (s5, 0.5), (s6, 0.2, 0.3)}
C41 LH = {(s2, 0.2), (s3, 0.4), (s4, 0.4, 0.5)} LH = {(s3, 0.3), (s4, 0.4, 0.5), (s5, 0.3)} LH = {(s3, 0.4), (s4, 0.5, 0.6), (s5, 0.2)} LH = {(s3, 0.4), (s4, 0.5, 0.6), (s5, 0.3)}
C42 LH = {(s1, 0.1, 0.2), (s3, 0.4, 0.5), (s4, 0.5)} LH = {(s1, 0.2), (s3, 0.4, 0.5), (s5, 0.5)} LH = {(s4, 0.4), (s5, 0.6), (s6, 0.1, 0.2)} LH = {(s4, 0.5), (s5, 0.2, 0.3)}
C43 LH = {(s2, 0.2, 0.3), (s3, 0.4), (s4, 0.4)} LH = {(s2, 0.3, 0.4), (s3, 0.4), (s5, 0.3, 0.4)} LH = {(s3, 0.3), (s2, 0.2), (s4, 0.3)} LH = {(s2, 0.3, 0.4), (s5, 0.3), (s6, 0.4)}
C51 LH = {(s2, 0.3), (s4, 0.3, 0.4), (s4, 0.2, 0.3)} LH = {(s2, 0.3), (s3, 0.4), (s4, 0.3)} LH = {(s4, 0.3, 0.4), (s5, 0.4)} LH = {(s3, 0.3, 0.4), (s4, 0.4), (s6, 0.1, 0.2)}
C52 LH = {(s3, 0.4), (s4, 0.2, 0.3), (s5, 0.3)} LH = {(s3, 0.2, 0.3), (s4, 0.4), (s5, 0.3, 0.4)} LH = {(s5, 0.5, 0.7), (s6, 0.2)} LH = {(s1, 0.4), (s2, 0.4, 0.5), (s4, 0.3)}
C53 LH = {(s2, 0.3), (s3, 0.4), (s4, 0.4, 0.5)} LH = {(s3, 0.2), (s4, 0.3, 0.4), (s5, 0.4)} LH = {(s4, 0.4, 0.5), (s5, 0.3, 0.4)} LH = {(s1, 0.3), (s2, 0.4), (s5, 0.4, 0.5)}
C54 LH = {(s2, 0.4), (s4, 0.5, 0.6)} LH = {(s4, 0.3, 0.4), (s5, 0.5), (s6, 0.1, 0.2)} LH = {(s3, 0.4), (s4, 0.3, 0.4), (s5, 0.1, 0.2)} LH = {(s4, 0.5, 0.6), (s6, 0.1, 0.2)}
C55 LH = {(s5, 0.5)} LH = {(s5, 0.5)} LH = {(s5, 0.5)} LH = {(s5, 0.5)}
Journal of Energy Storage 32 (2020) 101753
Table 17
Criteria risk values in 4-dimention for S-PHS project under the DBFO mode.
Y. Wu, et al.

RPM RPO RCS RUC

C11 LH = {(s3, 0.2, 0.3), (s4, 0.5), (s5, 0.3)} LH = {(s3, 0.2), (s4, 0.5), (s5, 0.4, 0.3)} LH = {(s4, 0.5), (s5, 0.4)} LH = {(s4, 0.4, 0.3), (s5, 0.6)}
C12 LH = {(s2, 0.3), (s4, 0.4, 0.6), (s5, 0.3)} LH = {(s4, 0.3), (s5, 0.5, 0.6), (s6, 0.4)} LH = {(s1, 0.2), (s3, 0.5), (s4, 0.3, 0.4)} LH = {(s2, 0.4), (s3, 0.2, 0.3), (s4, 0.1)}
C13 LH = {(s3, 0.3, 0.4), (s4, 0.6), (s5, 0.2, 0.3)} LH = {(s4, 0.3), (s5, 0.6), (s6, 0.5)} LH = {(s4, 0.4), (s5, 0.55, 0.65), (s5, 0.3)} LH = {(s2, 0.3, 0.4), (s3, 0.5), (s6, 0.1)}
C14 LH = {(s4, 0.4), (s5, 0.4), (s6, 0.2)} LH = {(s2, 0.1), (s4, 0.2), (s5, 0.7)} LH = {(s3, 0.3), (s4, 0.5), (s5, 0.4)} LH = {(s2, 0.4), (s4, 0.4, 0.5)}
C21 LH = {(s4, 0.5), (s5, 0.5, 0.7)} LH = {(s2, 0.3), (s5, 0.6), (s6, 0.2)} LH = {(s2, 0.2, 0.3), (s3, 0.3), (s4, 0.4)} LH = {(s3, 0.1), (s5, 0.5), (s6, 0.4)}
C22 LH = {(s4, 0.4), (s5, 0.6)} LH = {(s3, 0.6), (s5, 0.4, 0.5)} LH = {(s5, 0.5, 0.7), (s6, 0.1)} LH = {(s4, 0.4), (s5, 0.7, 0.8), (s6, 0.2)}
C23 LH = {(s2, 0.2, 0.3), (s4, 0.5), (s5, 0.3, 0.4)} LH = {(s4, 0.3), (s5, 0.5), (s6, 0.4)} LH = {(s2, 0.4), (s3, 0.4), (s4, 0.2, 0.4)} LH = {(s3, 0.4, 0.5), (s4, 0.2), (s5, 0.1, 0.2)}
C24 LH = {(s4, 0.3, 0.4), (s5, 0.5), (s6, 0.2, 0.3)} LH = {(s3, 0.2), (s4, 0.5), (s5, 0.4)} LH = {(s3, 0.2, 0.3), (s4, 0.7)} LH = {(s2, 0.3, 0.4), (s3, 0.5), (s4, 0.2)}
C25 LH = {(s5, 0.4), (s6, 0.5, 0.6)} LH = {(s2, 0.2), (s3, 0.6), (s4, 0.2)} LH = {(s3, 0.3), (s4, 0.4), (s6, 0.1, 0.2)} LH = {(s5, 0.6), (s4, 0.2), (s6, 0.3)}
C26 LH = {(s3, 0.3), (s4, 0.4, 0.6), (s5, 0.2)} LH = {(s4, 0.4), (s5, 0.5), (s6, 0.2)} LH = {(s2, 0.3), (s4, 0.4, 0.3), (s5, 0.3)} LH = {(s2, 0.3), (s5, 0.4), (s6, 0.1)}
C31 LH = {(s3, 0.3), (s4, 0.6, 0.7)} LH = {(s3, 0.2), (s4, 0.4), (s5, 0.5)} LH = {(s3, 0.4), (s4, 0.5), (s5, 0.1)} LH = {(s3, 0.4), (s4, 0.5), (s5, 0.4)}
C41 LH = {(s2, 0.2), (s3, 0.4, 0.5), (s4, 0.5)} LH = {(s1, 0.2), (s3, 0.4), (s5, 0.6)} LH = {(s4, 0.4), (s5, 0.6), (s6, 0.1, 0.2)} LH = {(s4, 0.5), (s5, 0.2, 0.3)}
C42 LH = {(s2, 0.2, 0.3), (s3, 0.4), (s4, 0.4)} LH = {(s2, 0.3, 0.4), (s3, 0.4), (s5, 0.3, 0.4)} LH = {(s3, 0.3), (s2, 0.2), (s4, 0.3)} LH = {(s2, 0.3, 0.4), (s4, 0.3), (s5, 0.4)}
C51 LH = {(s4, 0.3, 0.4), (s3, 0.2), (s5, 0.4)} LH = {(s2, 0.3), (s3, 0.3), (s4, 0.6)} LH = {(s4, 0.3, 0.4), (s5, 0.6)} LH = {(s3, 0.3, 0.4), (s4, 0.4), (s6, 0.1, 0.2)}
C52 LH = {(s3, 0.4), (s4, 0.2, 0.3), (s5, 0.3)} LH = {(s3, 0.2), (s4, 0.3), (s5, 0.4, 0.5)} LH = {(s5, 0.5, 0.6), (s4, 0.3), (s6, 0.1)} LH = {(s1, 0.4), (s2, 0.4), (s4, 0.2)}
C53 LH = {(s3, 0.3), (s4, 0.4), (s5, 0.3)} LH = {(s6, 0.1), (s4, 0.2, 0.4), (s5, 0.5)} LH = {(s4, 0.6), (s5, 0.3, 0.4)} LH = {(s3, 0.4), (s5, 0.6)}
C54 LH = {(s2, 0.4), (s4, 0.5, 0.6)} LH = {(s4, 0.3, 0.4), (s5, 0.5), (s6, 0.1, 0.2)} LH = {(s3, 0.4), (s4, 0.3, 0.4), (s5, 0.1, 0.2)} LH = {(s4, 0.5, 0.6), (s6, 0.1, 0.2)}
C55 LH = {(s5, 0.5)} LH = {(s5, 0.5)} LH = {(s5, 0.5)} LH = {(s5, 0.5)}

18
Table 18
Criteria risk values in 4-dimention for S-PHS project under the BOO mode.
RPM RPO RCS RUC

C11 LH = {(s2, 0.2, 0.3), (s4, 0.6), (s5, 0.3)} LH = {(s3, 0.3, 0.4), (s4, 0.5), (s5, 0.2, 0.3)} LH = {(s4, 0.4), (s5, 0.6)} LH = {(s4, 0.4, 0.5), (s5, 0.6)}
C12 LH = {(s2, 0.3), (s4, 0.5, 0.6), (s5, 0.2)} LH = {(s4, 0.2), (s5, 0.5, 0.6), (s6, 0.5)} LH = {(s1, 0.2), (s3, 0.4), (s4, 0.5)} LH = {(s2, 0.5), (s3, 0.2, 0.3), (s4, 0.3)}
C13 LH = {(s3, 0.3, 0.4), (s4, 0.6), (s5, 0.2, 0.3)} LH = {(s4, 0.3), (s5, 0.5, 0.7), (s6, 0.4)} LH = {(s4, 0.4), (s5, 0.55, 0.65), (s5, 0.1, 0.2)} LH = {(s2, 0.3, 0.4), (s3, 0.5), (s6, 0.2)}
C21 LH = {(s3, 0.2, 0.4), (s5, 0.5)} LH = {(s2, 0.1), (s4, 0.3), (s5, 0.7)} LH = {(s3, 0.3), (s4, 0.3, 0.2), (s5, 0.5)} LH = {(s2, 0.2), (s4, 0.3, 0.4), (s5, 0.4)}
C22 LH = {(s4, 0.3, 0.4), (s5, 0.7)} LH = {(s2, 0.2), (s4, 0.4, 0.5), (s5, 0.4)} LH = {(s2, 0.4), (s3, 0.2, 0.3), (s4, 0.3)} LH = {(s3, 0.2), (s4, 0.3, 0.4), (s5, 0.6)}
C23 LH = {(s4, 0.3, 0.4), (s5, 0.6), (s6, 0.1, 0.2)} LH = {(s3, 0.4), (s5, 0.5, 0.6)} LH = {(s4, 0.3, 0.4), (s5, 0.5, 0.7)} LH = {(s4, 0.4), (s5, 0.6, 0.7), (s6, 0.1, 0.2)}
C24 LH = {(s2, 0.2, 0.3), (s4, 0.5), (s5, 0.3, 0.4)} LH = {(s4, 0.3, 0.4), (s5, 0.5), (s6, 0.3)} LH = {(s2, 0.4), (s3, 0.4), (s4, 0.2, 0.3)} LH = {(s3, 0.4, 0.5), (s4, 0.2), (s5, 0.1, 0.2)}
C25 LH = {(s4, 0.3, 0.4), (s5, 0.5), (s6, 0.3)} LH = {(s3, 0.3), (s4, 0.5), (s5, 0.3, 0.4)} LH = {(s3, 0.2, 0.3), (s4, 0.7)} LH = {(s2, 0.3, 0.4), (s3, 0.5), (s4, 0.4)}
C31 LH = {(s4, 0.2), (s5, 0.5, 0.6), (s6, 0.3, 0.4)} LH = {(s3, 0.2), (s4, 0.6), (s5, 0.3, 0.4)} LH = {(s4, 0.3), (s5, 0.4), (s6, 0.1)} LH = {(s5, 0.5), (s4, 0.3, 0.4), (s6, 0.1)}
C32 LH = {(s2, 0.2), (s4, 0.4, 0.5)} LH = {(s4, 0.3, 0.4), (s5, 0.4, 0.5), (s6, 0.1)} LH = {(s3, 0.2), (s5, 0.4, 0.5), (s6, 0, 0.1)} LH = {(s2, 0.5), (s5, 0.4, 0.5), (s6, 0.2, 0.3)}
C41 LH = {(s2, 0.4), (s3, 0.3), (s4, 0.2)} LH = {(s3, 0.2), (s4, 0.3, 0.4), (s5, 0.5)} LH = {(s3, 0.5, 0.6), (s4, 0.3, 0.4), (s5, 0.2)} LH = {(s3, 0.2, 0.3), (s4, 0.4), (s5, 0.3, 0.4)}
C42 LH = {(s3, 0.3), (s5, 0.4, 0.5), (s6, 0.2, 0.3)} LH = {(s1, 0.4), (s3, 0.4, 0.5), (s5, 0.3)} LH = {(s4, 0.4), (s5, 0.6), (s6, 0.1, 0.2)} LH = {(s4, 0.5), (s5, 0.2, 0.3)}
C43 LH = {(s2, 0.2), (s4, 0.4, 0.5), (s5, 0.3)} LH = {(s4, 0.4), (s5, 0.5), (s6, 0.1, 0.2)} LH = {(s2, 0.2), (s4, 0.4, 0.5), (s5, 0.4)} LH = {(s2, 0.4), (s5, 0.5), (s6, 0.2, 0.3)}
C51 LH = {(s4, 0.3, 0.4), (s2, 0.3), (s5, 0.2, 0.3)} LH = {(s2, 0.3), (s3, 0.4), (s4, 0.3)} LH = {(s4, 0.3, 0.4), (s5, 0.4)} LH = {(s3, 0.3, 0.4), (s4, 0.4), (s6, 0.1, 0.2)}
C52 LH = {(s3, 0.4, 0.5), (s4, 0.2, 0.3), (s5, 0.2)} LH = {(s3, 0.2, 0.3), (s4, 0.4), (s5, 0.2)} LH = {(s5, 0.4), (s4, 0.5, 0.6)} LH = {(s1, 0.4), (s2, 0.3, 0.4), (s4, 0.3)}
C53 LH = {(s2, 0.3), (s3, 0.4), (s4, 0.4, 0.5)} LH = {(s3, 0.2), (s4, 0.4, 0.5), (s5, 0.3)} LH = {(s3, 0.3, 0.4), (s4, 0.5), (s5, 0.2, 0.3)} LH = {(s1, 0.3), (s2, 0.4), (s5, 0.2, 0.4)}
C54 LH = {(s2, 0.4), (s4, 0.5, 0.6)} LH = {(s4, 0.3, 0.4), (s5, 0.5), (s6, 0.1, 0.2)} LH = {(s3, 0.4, 0.5), (s4, 0.2, 0.3), (s5, 0.1, 0.2)} LH = {(s4, 0.5, 0.6), (s6, 0.3, 0.4)}
C55 LH = {(s5, 0.5)} LH = {(s5, 0.5)} LH = {(s5, 0.5)} LH = {(s5, 0.5)}
Journal of Energy Storage 32 (2020) 101753
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Appendix B

Table 19
BOT-S-PHS project cloud value of each criteria.
Criteria RPM RPO RCS RUC Aggregated cloud

C11 (5.94, 0.38, 0.35) (6.00, 0.34, 0.32) (6.89, 0.45, 0.31) (6.84, 0.44, 0.31) (6.40, 0.40, 0.33)
C12 (5.70, 0.37, 0.35) (7.93, 0.61, 0.52) (4.83, 0.29, 0.32) (4.55, 0.28, 0.25) (5.61, 0.37, 0.35)
C13 (6.00, 0.34, 0.32) (7.91, 0.60, 0.52) (7.27, 0.52, 0.52) (5.63, 0.44, 0.46) (6.64, 0.47, 0.46)
C21 (6.56, 0.42, 0.28) (6.77, 0.46, 0.35) (6.18, 0.37, 0.32) (5.58, 0.36, 0.35) (6.25, 0.40, 0.33)
C22 (6.69, 0.42, 0.31) (5.46, 0.35, 0.35) (5.25, 0.29, 0.25) (6.48, 0.40, 0.32) (5.94, 0.37, 0.31)
C23 (7.35, 0.52, 0.52) (6.44, 0.41, 0.28) (7.50, 0.52, 0.26) (7.30, 0.52, 0.52) (7.14, 0.49, 0.41)
C24 (6.01, 0.40, 0.35) (7.68, 0.58, 0.52) (4.86, 0.27, 0.25) (5.71, 0.31, 0.32) (5.98, 0.38, 0.35)
C25 (7.58, 0.56, 0.52) (6.18, 0.37, 0.32) (5.70, 0.29, 0.19) (5.03, 0.27, 0.25) (6.06, 0.36, 0.32)
C31 (8.01, 0.61, 0.52) (4.90, 0.24, 0.25) (6.33, 0.43, 0.46) (7.83, 0.59, 0.52) (6.65, 0.45, 0.43)
C32 (6.04, 0.40, 0.35) (7.73, 0.52, 0.52) (6.18, 0.41, 0.35) (6.84, 0.58, 0.52) (6.56, 0.47, 0.44)
C41 (5.23, 0.27, 0.25) (6.12, 0.36, 0.32) (5.89, 0.33, 0.32) (6.02, 0.34, 0.32) (5.81, 0.33, 0.30)
C42 (4.98, 0.29, 0.32) (5.26, 0.37, 0.38) (7.29, 0.52, 0.52) (6.47, 0.39, 0.31) (5.93, 0.39, 0.39)
C43 (5.14, 0.27, 0.25) (5.49, 0.36, 0.32) (5.12, 0.27, 0.25) (5.91, 0.41, 0.35) (5.40, 0.33, 0.29)
C51 (5.75, 0.39, 0.35) (5, 0.27, 0.25) (6.78, 0.43, 0.31) (5.84, 0.32, 0.46) (5.81, 0.35, 0.35)
C52 (6.04, 0.35, 0.32) (6.26, 0.38, 0.32) (8.13, 0.61, 0.50) (4.01, 0.36, 0.35) (5.92, 0.43, 0.39)
C53 (5.12, 0.28, 0.25) (6.40, 0.40, 0.32) (6.63, 0.42, 0.31) (4.99, 0.46, 0.40) (5.74, 0.40, 0.34)
C54 (5.15, 0.32, 0.23) (7.33, 0.52, 0.52) (5.79, 0.31, 0.32) (6.82, 0.48, 0.45) (6.22, 0.40, 0.38)
C55 (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26)

Table 20
DBFO-S-PHS project risk level of each criteria.
Criteria RPM RPO RCS RUC Aggregated cloud

C11 (6.17, 0.36, 0.32) (6.29, 0.38, 0.32) (6.64, 0.42, 0.31) (6.93, 0.45, 0.31) (6.50, 0.40, 0.32)
C12 (5.86, 0.39, 0.35) (7.93, 0.61, 0.52) (4.84, 0.28, 0.32) (4.62, 0.27, 0.25) (5.68, 0.37, 0.35)
C13 (6.00, 0.34, 0.32) (8.06, 0.62, 0.52) (7.61, 0.57, 0.52) (5.17, 0.36, 0.46) (6.61, 0.46, 0.47)
C14 (7.83, 0.54, 0.52) (6.85, 0.47, 0.35) (6.23, 0.37, 0.32) (5.06, 0.32, 0.23) (6.32, 0.42, 0.35)
C21 (6.80, 0.44, 0.31) (7.01, 0.57, 0.52) (5.15, 0.28, 0.25) (8.25, 0.40, 0.32) (7.08, 0.44, 0.37)
C22 (6.88, 0.45, 0.31) (6.07, 0.37, 0.28) (7.86, 0.57, 0.50) (7.41, 0.53, 0.52) (7.02, 0.48, 0.40)
C23 (6.01, 0.40, 0.35) (7.95, 0.61, 0.52) (4.91, 0.27, 0.25) (5.71, 0.31, 0.32) (6.05, 0.39, 0.35)
C24 (7.58, 0.56, 0.52) (6.34, 0.39, 0.32) (5.70, 0.29, 0.19) (4.86, 0.26, 0.25) (6.04, 0.37, 0.32)
C25 (8.95, 0.72, 0.50) (5, 0.25, 0.25) (6.33, 0.43, 0.46) (7.90, 0.60, 0.52) (6.89, 0.48, 0.43)
C26 (5.98, 0.34, 0.32) (7.39, 0.54, 0.52) (5.84, 0.40, 0.35) (6.52, 0.53, 0.52) (6.40, 0.45, 0.43)
C31 (5.65, 0.28, 0.19) (6.48, 0.40, 0.32) (5.72, 0.30, 0.32) (6.14, 0.36, 0.32) (5.99, 0.34, 0.29)
C41 (5.25, 0.27, 0.25) (5.83, 0.42, 0.38) (7.29, 0.52, 0.52) (6.47, 0.39, 0.31) (6.16, 0.40, 0.37)
C42 (5.14, 0.27, 0.25) (5.49, 0.36, 0.32) (5.12, 0.27, 0.25) (5.91, 0.41, 0.35) (5.40, 0.33, 0.29)
C51 (6.40, 0.39, 0.32) (3.75, 0.28, 0.25) (6.93, 0.45, 0.31) (5.84, 0.32, 0.46) (5.58, 0.36, 0.35)
C52 (6.04, 0.35, 0.32) (6.49, 0.41, 0.32) (7.26, 0.51, 0.52) (3.81, 0.37, 0.35) (5.74, 0.42, 0.40)
C53 (6.31, 0.36, 0.32) (7.26, 0.51, 0.52) (6.53, 0.40, 0.31) (6.50, 0.42, 0.28) (6.64, 0.42, 0.36)
C54 (5.15, 0.32, 0.23) (7.33, 0.52, 0.52) (5.79, 0.31, 0.32) (6.82, 0.48, 0.45) (6.22, 0.40, 0.38)
C55 (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26)

Table 21
BOO-S-PHS project cloud value of each criteria.
Criteria RPM RPO RCS RUC Aggregated cloud

C11 (5.94, 0.38, 0.35) (6.00, 0.34, 0.32) (6.89, 0.45, 0.31) (6.84, 0.44, 0.31) (6.40, 0.40, 0.33)
C12 (5.70, 0.37, 0.35) (8.25, 0.65, 0.52) (4.83, 0.29, 0.32) (4.82, 0.29, 0.25) (5.75, 0.38, 0.35)
C13 (6.00, 0.34, 0.32) (7.91, 0.60, 0.52) (7.27, 0.52, 0.52) (5.63, 0.44, 0.46) (6.64, 0.47, 0.46)
C21 (7.25, 0.54, 0.51) (6.42, 0.45, 0.35) (6.42, 0.40, 0.32) (6.20, 0.42, 0.35) (6.56, 0.45, 0.38)
C22 (7.00, 0.46, 0.31) (6.18, 0.41, 0.35) (4.90, 0.29, 0.25) (6.60, 0.42, 0.32) (6.11, 0.39, 0.31)
C23 (7.35, 0.52, 0.52) (6.44, 0.41, 0.28) (6.93, 0.45, 0.31) (7.30, 0.52, 0.52) (7.0, 0.48, 0.42)
C24 (6.01, 0.40, 0.35) (7.68, 0.58, 0.52) (4.86, 0.27, 0.25) (5.71, 0.31, 0.32) (5.98, 0.38, 0.35)
C25 (7.68, 0.58, 0.52) (6.18, 0.37, 0.32) (5.70, 0.29, 0.19) (5.03, 0.27, 0.25) (6.08, 0.37, 0.32)
C31 (8.01, 0.61, 0.52) (6.26, 0.37, 0.32) (7.23, 0.51, 0.52) (7.19, 0.50, 0.52) (7.15, 0.50, 0.47)
C32 (5.37, 0.32, 0.23) (7.18, 0.50, 0.52) (7.17, 0.52, 0.51) (6.58, 0.56, 0.52) (6.53, 0.47, 0.44)
C41 (4.79, 0.28, 0.25) (6.51, 0.41, 0.32) (5.76, 0.31, 0.32) (6.26, 0.38, 0.32) (5.79, 0.34, 0.30)
C42 (7.38, 0.56, 0.51) (4.78, 0.38, 0.38) (7.29, 0.52, 0.52) (6.47, 0.39, 0.31) (6.39, 0.46, 0.44)
C43 (6.04, 0.40, 0.35) (7.27, 0.52, 0.52) (6.18, 0.41, 0.35) (6.84, 0.58, 0.52) (6.56, 0.47, 0.44)
C51 (5.75, 0.39, 0.35) (5, 0.27, 0.25) (6.78, 0.43, 0.31) (5.84, 0.32, 0.46) (5.81, 0.35, 0.35)
C52 (5.82, 0.32, 0.32) (6.04, 0.35, 0.32) (6.61, 0.41, 0.31) (4.00, 0.36, 0.35) (5.52, 0.38, 0.35)
C53 (5.12, 0.28, 0.25) (6.24, 0.37, 0.32) (6.00, 0.34, 0.32) (4.62, 0.46, 0.40) (5.46, 0.38, 0.34)
C54 (5.15, 0.32, 0.23) (7.33, 0.52, 0.52) (5.72, 0.31, 0.32) (6.56, 0.41, 0.31) (6.14, 0.38, 0.34)
C55 (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26) (7.5, 0.52, 0.26)

19
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

Appendix C

Table 22
The T(Aαi) and βi of S-PHS project under BOT mode.
Criteria T(Aαi) βi

C11 16.044 0.0561


C12 16.211 0.0567
C13 15.132 0.0532
C21 16.123 0.0564
C22 15.865 0.0555
C23 15.667 0.0549
C24 16.215 0.0567
C25 15.780 0.0553
C31 15.645 0.0548
C32 15.362 0.0539
C41 15.253 0.0536
C42 16.131 0.0565
C43 15.541 0.0545
C51 16.112 0.0564
C52 15.897 0.0557
C53 16.240 0.0568
C54 16.153 0.0565
C55 16.066 0.0562

Table 23
The T(Aαi) and βi of S-PHS project under DBFO mode.
Criteria T(Aαi) βi

C11 15.875 0.0555


C12 16.295 0.0569
C13 15.172 0.0532
C14 16.251 0.0568
C21 15.938 0.0557
C22 15.844 0.0554
C23 16.194 0.0566
C24 15.734 0.0551
C25 15.619 0.0547
C26 15.619 0.0547
C31 15.053 0.0528
C41 16.228 0.0567
C42 15.470 0.0542
C51 16.177 0.0565
C52 15.847 0.0554
C53 16.226 0.0567
C54 16.192 0.0566
C55 16.139 0.0564

Table 24
T(Aαi) and βi of S-PHS project under BOO mode.
Criteria T(Aαi) βi

C11 15.892 0.0559


C12 16.117 0.0567
C13 15.404 0.0543
C21 15.919 0.0560
C22 15.723 0.0554
C23 15.774 0.0555
C24 16.090 0.0566
C25 15.590 0.0549
C31 15.349 0.0541
C32 15.511 0.0547
C41 15.258 0.0538
C42 16.598 0.0550
C43 15.587 0.0550
C51 15.961 0.0562
C52 16.161 0.0568
C53 16.163 0.0568
C54 16.004 0.0563
C55 15.928 0.0560

20
Y. Wu, et al. Journal of Energy Storage 32 (2020) 101753

References [33] H. Zhang, et al., Targeted poverty alleviation using photovoltaic power in China:
identifying financial options through a dynamic game analysis, Resour. Conserv.
Recycl. 139 (2018) 333–337.
[1] N.I. Ibrahim, F.A. Alsulaiman, F.N. Ani, Solar absorption systems with integrated [34] W. Hao, A study on the definition and classification of PPP, Urban Rapid Rail
absorption energy storage–A review, Renew. Sustain. Energy Rev. 82 (2018) Transit. 17 (5) (2004) 23–27.
1602–1610. [35] Q. Li, Y. Wang, Selection of public-private-partnership (PPP) modes for urban un­
[2] J.L. Crespo-Vazquez, et al., A machine learning based stochastic optimization fra­ derground utility tunnels, Tunnel Construct. 37 (9) (2017) 1077–1082.
mework for a wind and storage power plant participating in energy pool market, [36] Agency, I.E., Prospects for distributed energy systems in China. 2017.
Appl. Energy 232 (2018) 341–357. [37] J. Yuan, et al., Linguistic hesitant fuzzy multi-criterion decision-making for re­
[3] A. Parwal, et al., Energy management for a grid-connected wave energy park newable energy: a case study in Jilin, J. Clean. Prod. 172 (2018) 3201–3214.
through a hybrid energy storage system, Appl. Energy 231 (2018) 399–411. [38] C. Zhu, L. Zhu, X. Zhang, Linguistic hesitant fuzzy power aggregation operators and
[4] Port, P.o.L.B.T.G., Energy storage technologies white paper 2016. their applications in multiple attribute decision-making, Inform. Sci. 367 (2016)
[5] X. Xu, et al., Optimized sizing of a standalone PV-wind-hydropower station with 809–826.
pumped-storage installation hybrid energy system, Renew. Energy 147 (2020) [39] V. Torra, Hesitant fuzzy sets, Int. J. Intell. Syst. 25 (6) (2010) 529–539.
1418–1431. [40] F. Meng, X. Chen, Q. Zhang, Multi-attribute decision analysis under a linguistic
[6] M.S. Javed, et al., Solar and wind power generation systems with pumped hydro hesitant fuzzy environment, Inform. Sci. 267 (2014) 287–305.
storage: review and future perspectives, Renew. Energy 148 (2020) 176–192. [41] H. Liao, Z. Xu, X.-J. Zeng, Distance and similarity measures for hesitant fuzzy lin­
[7] E.N. Nyeche, E.O. Diemuodeke, Modelling and optimisation of a hybrid PV-wind guistic term sets and their application in multi-criteria decision making, Inform. Sci.
turbine-pumped hydro storage energy system for mini-grid application in coastline 271 (2014) 125–142.
communities, J. Cleaner Prod. 250 (2020) 119578. [42] Y. Wu, et al., A cloud decision framework in pure 2-tuple linguistic setting and its
[8] Y. Wu, et al., Optimal location selection for offshore wind-PV-seawater pumped application for low-speed wind farm site selection, J. Clean. Prod. 142 (2017)
storage power plant using a hybrid MCDM approach: a two-stage framework, 2154–2165.
Energy Convers. Manag. 199 (2019) 112066. [43] L. Zhang, et al., Renewable energy project performance evaluation using a hybrid
[9] A. Hiratsuka, T. Arai, T. Yoshimura, Seawater pumped-storage power plant in multi-criteria decision-making approach: case study in Fujian, China, J. Cleaner
Okinawa island, Japan, Eng. Geol. 35 (3) (1993) 237–246. Prod. 206 (2019) 1123–1137.
[10] C. Ioakimidis, K. Genikomsakis, Integration of seawater pumped-storage in the [44] D. Wang, et al., A cloud model-based approach for water quality assessment,
energy system of the Island of São Miguel (Azores), Sustainability 10 (10) (2018) Environ. Res. 148 (2016) 24–35.
3438. [45] A.J. Sangster, Massive energy storage systems enable secure electricity supply from
[11] Francisco, F.G.A., Wave energy distribution across the Agulhas Bank, a source of renewables, Modern Power Syst. 4 (4) (2016) 659–667.
renewable energy for a seawater pumped storage scheme. 2013. [46] M. Tian, et al., Risk and profit-based bidding and offering strategies for pumped
[12] L. Bridier, M. David, P. Lauret, Optimal design of a storage system coupled with hydro storage in the energy market, J. Clean. Prod. 256 (2020) 120715.
intermittent renewables, Renew. Energy 67 (4) (2014) 2–9. [47] Z. Lu, Y. Gao, W. Zhao, A TODIM-based approach for environmental impact as­
[13] N. Ghorbani, H. Makian, C. Breyer, A GIS-based method to identify potential sites sessment of pumped hydro energy storage plant, J. Clean. Prod. 248 (2020) 119265.
for pumped hydro energy storage - Case of Iran, Energy 169 (2019) 854–867. [48] X. Jiang, et al., Identifying significant risks and analyzing risk relationship for
[14] N. Liang, et al., Two-stage coordinate optimal scheduling of seawater pumped construction PPP projects in China using integrated FISM-MICMAC approach,
storage in active distribution networks, Sustainability 10 (6) (2018) 2014. Sustainability 11 (19) (2019) 5206.
[15] NEA, The results of resource census of seawater pumped storage power station [49] D.A. Katsaprakakis, et al., Technical details regarding the design, the construction
released by National Energy Administrationhttp://www.2017-04/05/contengov. and the operation of seawater pumped storage systems, Energy 55 (2013) 619–630.
cn/xinwen/t_5183621.htm, 2017(68). [50] G. Manfrida, R. Secchi, Seawater pumping as an electricity storage solution for
[16] R. Segurado, et al., Optimization of a wind powered desalination and pumped hydro photovoltaic energy systems, Energy 69 (2014) 470–484.
storage system, Appl. Energy 177 (2016) 487–499. [51] Y. Wu, et al., Risk assessment in straw-based power generation public-private
[17] M. Bertsiou, et al., Water management and electricity output of a Hybrid Renewable partnership projects in China: a fuzzy synthetic evaluation analysis, J. Clean. Prod.
Energy System (HRES) in Fournoi Island in Aegean Sea, Renew. Energy 118 (2018) 161 (2017) 977–990.
790–798. [52] K.Lewis, D.G.M., Public private partnerships: the worldwide revolution in infra­
[18] Y. Wu, et al., Location selection of seawater pumped hydro storage station in China structure provision and project finance.
based on multi-attribute decision making, Renew. Energy 139 (2019) 410–425. [53] J. Sang, et al., Design of build-operate-transfer contract for integrated rail and
[19] U. Portero, S. Velázquez, J.A. Carta, Sizing of a wind-hydro system using a re­ property development with uncertainty in future urban population, Transp. Res.
versible hydraulic facility with seawater. A case study in the Canary Islands, Energy Part B-Methodol. 130 (2019) 36–66.
Convers. Manag. 106 (2015) 1251–1263. [54] L. Kumar, A. Jindal, N.R. Velaga, Financial risk assessment and modelling of PPP
[20] A. Sokołowski, D. Brulińska, The effects of low seawater pH on energy storage and based Indian highway infrastructure projects, Transp. Policy 62 (2018) 2–11.
heat shock protein 70 expression in a bivalve Limecola balthica, Mar. Environ. Res. [55] J. Song, et al., Risk identification for PPP waste-to-energy incineration projects in
140 (2018) 289–298. China, Energy Policy 61 (2013) 953–962.
[21] Z. Wu, et al., Mathematical modeling and performance analysis of seawater heat [56] J. Liu, Q. Wei, Risk evaluation of electric vehicle charging infrastructure public-
exchanger in closed-loop seawater-source heat pump system, J. Energy Eng.-ASCE private partnership projects in China using fuzzy TOPSIS, J. Clean. Prod. 189
145 (4) (2019) 04019012. (2018) 211–222.
[22] E. McLean, D. Kearney, An evaluation of seawater pumped hydro storage for reg­ [57] X. Zhao, B.-G. Hwang, Y. Gao, A fuzzy synthetic evaluation approach for risk as­
ulating the export of renewable energy to the national grid, Energy Procedia 46 sessment: a case of Singapore's green projects, J. Clean. Prod. 115 (2016) 203–213.
(2014) 152–160. [58] Kang Jia, J.S., Public private partnerships theory and practice2015.
[23] L. Kumar, A. Jindal, N.R. Velaga, Financial risk assessment and modelling of PPP [59] X. Gou, Z. Xu, H. Liao, Multiple criteria decision making based on Bonferroni means
based Indian highway infrastructure projects, Transp. Policy 62 (2017). with hesitant fuzzy linguistic information, Soft Computing, (2017).
[24] Y. Li, X. Wang, Risk assessment for public–private partnership projects: using a [60] W. Song, J. Zhu, A multistage risk decision making method for normal cloud model
fuzzy analytic hierarchical process method and expert opinion in China, J. Risk Res. considering behavior characteristics, Appl. Soft Comput. 78 (2019) 393–406.
21 (8) (2018) 952–973. [61] Y. Wu, et al., Cloud-based decision framework for waste-to-energy plant site se­
[25] Sokołowski, A. and D. Brulińska, The effects of low seawater pH on energy storage lection - a case study from China, Waste Manag. (Oxford) 48 (48) (2016) 593–603.
and heat shock protein 70 expression in a bivalve Limecola balthica. Mar. Environ. [62] H. Wang, Z. Xu, X. Zeng, Hesitant fuzzy linguistic term sets for linguistic decision
Res., 2018. making: current developments, issues and challenges, Inform. Fusion 43 (2018)
[26] Y. Wu, et al., Risk assessment on offshore photovoltaic power generation projects in 1–12.
China based on a fuzzy analysis framework, J. Clean. Prod. 215 (2019) 46–62. [63] H.-l Wang, Y.-q. Feng, On multiple attribute group decision making with linguistic
[27] Y. Wu, et al., Renewable energy investment risk assessment for nations along assessment information based on cloud model, Control Decis. 20 (6) (2005) 679.
China's Belt & Road Initiative: an ANP-cloud model method, Energy 190 (2020) [64] Hu, S., Multi-attribute group decision making based on comprehensive cloud under
116381. linguistic hesitant fuzzy environment and its application in ERP selection. Jiangxi
[28] T. Yang, et al., Application of the public–private partnership model to urban sewage Univ. Fin. Econom.. 2018.
treatment, J. Clean. Prod. 142 (2017) 1065–1074. [65] W. Jun, Z. Jianjun, L. Xiaodi, An integrated similarity measure method for normal
[29] S. Geng, et al., Social capital selection framework of public-private partnership cloud model based on shape and distance, Syst. Eng. Theory Pract. 37 (3) (2017)
project of electrochemical storage power station under linguistic environment, J. 742–751.
Clean. Prod. 199 (2018) 751–762. [66] Y. Wu, et al., Risk management of public-private partnership charging infra­
[30] M. Buso, A. Stenger, Public-private partnerships as a policy response to climate structure projects in China based on a three-dimension framework, Energy 165
change, Energy Policy 119 (2018) 487–494. (2018) 1089–1101.
[31] Y. Wu, et al., A risk assessment framework of PPP waste-to-energy incineration [67] A.A. Ahmadabadi, G. Heravi, Risk assessment framework of PPP-megaprojects fo­
projects in China under 2-dimension linguistic environment, J. Cleaner Prod. 183 cusing on risk interaction and project success, Transp. Res. Part A-Policy Pract. 124
(2018) 602–617. (2019) 169–188.
[32] Y. Xu, et al., Critical risk factors affecting the implementation of PPP waste-to-
energy projects in China, Appl. Energy 158 (2015) 403–411.

21

You might also like