Professional Documents
Culture Documents
Zhao Et Al. (202) Economy-Environment-Energy Performance Evaluation of CCHP Microgrid System - A Hybrid Multi-Criteria Decision-Making Method
Zhao Et Al. (202) Economy-Environment-Energy Performance Evaluation of CCHP Microgrid System - A Hybrid Multi-Criteria Decision-Making Method
Zhao Et Al. (202) Economy-Environment-Energy Performance Evaluation of CCHP Microgrid System - A Hybrid Multi-Criteria Decision-Making Method
Energy
journal homepage: www.elsevier.com/locate/energy
a r t i c l e i n f o a b s t r a c t
Article history: The combined cooling, heating and power micro-grid (CCHP-MG) system has significant economy-
Received 30 April 2021 environment-energy (3E) characteristics, making it valuable to evaluating its 3E performance in pro-
Received in revised form moting the 3E coordinated development. Based on the energy supply structure of CCHP-MG system, this
24 November 2021
paper constructs a 3E performance evaluation index system for the CCHP-MG system from economy,
Accepted 5 December 2021
Available online 7 December 2021
environment and energy dimensions. Then, a hybrid multi-criteria decision-making (MCDM) framework
is designed, involving integrated weighting method based on anti-entropy weight method and grey
decision-making trial and evaluation laboratory (grey-DEMATEL), and improved technique for order
Keywords:
3E performance
preference by similarity to an ideal solution (TOPSIS) with difference and quotient grey relation analysis
CCHP micro-Grid system (DQGRA). 8 building-typed CCHP-MG systems are taken as examples for analysis. The results show that
Hybrid MCDM framework the benefits of system energy supply, equivalent emission reduction of pollutants, energy utilization and
Integrated subjective and objective equivalent energy saving are important manifestations of CCHP-MG's 3E performance, and raising energy
weighting supply scale and renewable energy proportion are the key ways to improve CCHP-MG's 3E performance.
Improved TOPSIS Comparing with 4 other MCDM models, the proposed hybrid MCDM model has good applicability and
Robustness and sample discrimination tests effectiveness for the 3E performance evaluation of CCHP-MG system, but it is necessary to carry out
targeted adjustment when applying it to other MCDM issues.
© 2021 Elsevier Ltd. All rights reserved.
https://doi.org/10.1016/j.energy.2021.122830
0360-5442/© 2021 Elsevier Ltd. All rights reserved.
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
high-grade energy of different energy sources for power generation CCHP residential system considering economic costs and pollution
and low-grade energy for heating and cooling, the energy efficiency emissions, showing that the CCHP residential system with envi-
of CCHP system can reach more than 80%, which is significantly ronmental and economic advantages should be promoted widely
better than that of traditional centralized power plants [13,14]. [38]. Zhao et al. established a modeling framework of a multi-
Distributed power generation and CCHP have outstanding ad- energy system (MES) with CCHP units, and examined the opera-
vantages, and their development has been widely promoted. tion strategy considering operation cost and carbon dioxide emis-
However, they also have many problems, such as the high cost of sion of the system, finding that the proposed operation strategy can
single-machine access to the grid and difficulty in control [15,16]. In improve the economic and environmental advantages of the sys-
order to give full play to the advantages of distributed power, tem [39]. However, in the research on the performance of the
realize the flexible application of distributed power, and solve the CCHP-MG system, previous researches mainly evaluated the per-
problem of large and diverse distributed power grid connection, formance of CCHP-MG system from the dimensions of technology,
the concept of micro-grid (MG), which is a controllable unit that economy, society and environment [40e42]. Although the indicator
integrates various distributed power sources, loads, energy storage system contains some indicators that reflect 3E characteristics, it
devices and energy conversion devices, is proposed [17,18]. MG can still does not deviate from the traditional evaluation framework for
provide cooling, heating, and electrical loads to the system by power transmission and distribution grid performance, making it
controlling the effective interconnection of distributed power and difficult to fully reflect the 3E characteristics of the CCHP-MG sys-
CCHP. In this case, the MG can be called a CCHP-MG system, and its tem, resulting that the evaluation results lack target reference for
definition can be extended to a micro energy grid, which is actually promoting 3E coordinated development of the CCHP-MG system.
a small IES [19,20]. The development and extension of the CCHP- In line with the above discussion, this paper examines the per-
MG can fully promote the large-scale access of distributed power formance evaluation of the CCHP-MG system from the perspective
sources and renewable energy sources, realize the highly reliable of 3E coordinated development, which can effectively make up for
supply of multiple energy forms, and promote the coordinated the insufficiency of previous research on CCHP-MG's 3E coordina-
development of the regional 3E system [21,22]. tion performance analysis. Based on the introduction of the energy
Early researches on the CCHP-MG mainly focused on the supply structure of the CCHP-MG system, the 3E performance
structure and energy supply mode of the CCHP-MG, with the pur- evaluation index system of the CCHP-MG system is constructed
pose of clarifying the basic concept and system framework of the from the three dimensions of economy, environment and energy.
CCHP-MG [23e26]. Subsequently, scholars’ researches on CCHP- Furthermore, a hybrid multi-criteria decision-making (MCDM)
MG mainly focus on its economic operation optimization. Specif- model based on the anti-entropy weight (AEW) method, the grey
ically, based on the CCHP-MG operation mode, corresponding decision-making trial and evaluation laboratory (grey-DEMATEL)
operating formulas for various power sources and devices in the and the improved technique for order preference by similarity to an
system are constructed, the objective function and constraint ideal solution (TOPSIS) with difference and quotient grey relation
conditions of the economic operation optimization problem of the analysis (DQGRA) is constructed (called DQGRA-TOPSIS), and
CCHP-MG system are designed accordingly, and the applicability several CCHP-MG systems are used for case analysis to verify the
and effectiveness of the model are verified through calculation applicability and effectiveness of the proposed model.
examples [27e29]. Li et al. developed an improved two-stage To sum up, the main contribution of this paper is twofold. First,
robust optimization model for CCHPeP2G microgrid system, by analyzing the 3E benefits of the CCHP-MG, this paper constructs
which can fully considers the multi-energy operation under wind an index system involving 13 indicators for evaluating the 3E per-
power outputs uncertainties [30]. Bu et al. examined the day-ahead formance of CCHP-MG system. Accordingly, based on grey-
optimization economic dispatch of CCHP multi-microgrid system, DEMATEL method, the interaction relationship between various
finding that the use of negotiation game in the scheduling process indicators is examined, which provides a reference for the 3E co-
helps to better realize the economics of the system [31]. Abdalla ordinated development of the CCHP-MG system. Second, a hybrid
et al. explored the optimized economic operation of the CCHP-MG MCDM model based on AEW-grey-DEMATEL and DQGRA-TOPSIS
with hybrid energy storage system, holding that the CCHP system techniques is developed, which has good applicability and effec-
can effectively reduce the operation and fuel costs of the MG [32]. tiveness for the evaluation issue in this paper. On one hand, the
In general, the above-mentioned researches on the CCHP-MG re- integrated weighting method based on AEW and grey-DEMATEL
veals the operation rules of the CCHP-MG system by modeling the techniques can deal with the interrelationship between the in-
equipment, and enriches the theories and practical outcomes of the dicators under the premise of reducing subjectivity, thereby
economic operation of the CCHP-MG, which provides a useful improving the reliability of the weighting results. On the other
reference for promoting the development of CCHP-MG and opti- hand, considering geometric similarity and numerical similarity
mizing its dispatching operation mode. between sequences, the DQGRA-TOPSIS technique is developed,
The CCHP-MG has significant 3E system characteristics, so it has which can overcome the vertical bisector problem of conventional
significant theoretical and practical value to examine the CCHP-MG TOPSIS using Euclidean distance as the closeness, and avoid the
operation mode and its performance from the perspective of 3E subjectivity in the determination of the resolution coefficient in
coordinated development [33]. Some scholars have considered conventional GRA.
environmental factors (environmental externality costs, pollutant The rest of this paper is organized as follow: Section 2 describes
discharge constraints, etc.) in addition to traditional economic and the energy supply structure of CCHP-MG system, and constructs the
energy factors when conducting research on the optimization of 3E performance evaluation index system of the CCHP-MG. Section 3
the economic operation of the CCHP-MG, which can be considered introduces the developed hybrid MCDM model for the 3E perfor-
as analyzing the operation mode of the CCHP-MG system from the mance evaluation of CCHP-MG, and discusses the applicability of
perspective of 3E coordinated development [34e36]. Guo et al. the proposed model. In Section 4, 8 building-typed CCHP-MG cases
examined the economic-environmental coordinated operation of are used for case analysis and the results are reported. Section 5
an industrial energy parks integrated with CCHP units, finding that investigates the interactions between the indicators, and verifies
under the proposed coordinated approach, the total operation cost the effectiveness of the model from the dimensions of robustness
and emission pollution are decreased significantly [37]. Yuan et al. and sample discrimination tests. Finally, Section 6 summarizes the
developed a novel algorithm to optimize the operation of a hybrid paper and put forward corresponding suggestions.
2
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
2. Energy supply structure and 3E performance of CCHP-MG and optimized, as well as coordinated and interacted, which fully
system reflects the 3E performance of the CCHP-MG system.
For economic performance, during the operation of CCHP-MG
2.1. Energy supply structure of CCHP-MG system system, its economic performance is mainly reflected in two as-
pects. One is the direct benefits, that is, the benefits obtained by
In order to achieve energy conservation, emission reduction, providing users with cooling, heating and power. The other is the
and improvement of energy utilization, the integrated energy indirect benefits, that is, the CCHP-MG can exchange energy with
supply technique in the form of MG including distributed energy the main power grid through the public connection points, serving
and CCHP systems, namely the CCHP-MG system, has received as backups for power system, thus can effectively save the con-
more and more attention [43]. Fig. 1 is a typical structure of the MG, struction investment related to improving the reliability of power
which is connected to the power system through isolation equip- supply [45]. Moreover, the CCHP-MG's “on-site production and on-
ment and isolation transformers. Most of the distributed power site consumption” mode can significantly reduce the grid loss in the
sources in the MG system use power electronic devices to connect power transmission and distribution process, thereby obtaining
to the load and the power system. significant grid loss reduction benefits.
Different from the typical MG system, the CCHP-MG not only For environmental performance, the CCHP-MG system contains
generates residual heat from the combined heat and power (CHP) gas generators and renewable energy sources such as wind and
equipment, but also contains a CCHP system to generate heat and solar power sources. Compared with traditional energy supply
cold power at the same time, so as to achieve efficient use of energy. modes, the CCHP-MG system can effectively promote the use of
Specifically, the CCHP-MG takes the micro gas turbine (MGT) typed clean energy by coordinating internal resources [41]. In the process
CCHP system as the core, and determines the structure, equipment of meeting the regional diversified energy demand, CCHP-MG
capacity, and power source output of the micro-energy grid ac- system will emit lower pollutants and reduce the damage to the
cording to the demand for cold, heat and electrical loads [44]. Fig. 2 environment, thus having significant emission reduction benefits.
shows the energy supply structure of a typical CCHP-MG system, For energy performance, the CCHP-MG system effectively sup-
including wind power generation, photovoltaic power generation, presses the instability of renewable energy output such as wind
CCHP systems, and electric and thermal energy storage devices. and solar through a variety of energy storage devices, and promotes
Among them, the CCHP system with MGT as the core includes MGT, the consumption of renewable energy [46]. Compared with tradi-
gas boilers and refrigerators, which recycles the residual heat tional energy supply modes, the CCHP-MG system can realize the
generated by the MGT to realize the combined supply of cooling, effective conversion of electricity, heating and cooling, and the
heating and electricity on the load side. In the CCHP-MG system, energy cascade utilization through energy storage devices, residual
photovoltaic power generation, wind power generation, MGT, heat recovery devices and energy conversion devices, thereby
electricity energy storage, and power purchase from the power grid significantly improving energy utilization efficiency.
are applied to meet the electrical load demand; MGT, gas-fired Based on the above analysis, this paper constructs an evaluation
boilers and thermal energy storage are applied to meet the heat index system reflecting the performance of the CCHP-MG from the
load demand; absorption chillers and electric chillers are applied to perspective of 3E framework, as shown in Fig. 3. According to Fig. 3,
meet the cooling load demand. Therefore, the CCHP-MG system can the economic performance of the CCHP-MG system mainly in-
improve energy utilization, reduce energy costs, improve energy cludes energy supply benefits, grid loss reduction benefits, and
supply reliability, and enhance environmental protection by energy investment savings benefits; the environmental performance
comprehensive and cascade utilization. mainly includes carbon emission reduction benefits and other
pollutant emission reduction benefits; the energy performance
2.2. Analysis on the 3E performance of CCHP-MG system mainly includes renewable energy consumption benefits, energy
utilization efficiency improvement benefits, and energy saving
The CCHP-MG system has significant 3E characteristics, and the benefits.
internal economic, environmental and energy factors are coupled
3
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
3. Hybrid evaluating model for the 3E performance of CCHP- indicator in the index system, and based on the judgment results,
MG system the direct grey relationship matrix of DEMATEL is constructed.
Furthermore, the direct grey relationship matrix is clarified and
3.1. Framework of the hybrid MCDM model realized, then the DEMATEL comprehensive influence matrix that
integrates the judgment results of multiple experts can be calcu-
The framework of the proposed hybrid MCDM model for 3E lated, and the subjective weight of the indicator based on grey-
performance evaluation of CCHP-MG system in this paper is shown DEMATEL is calculated accordingly. Finally, an optimization
in Fig. 4. The hybrid MCDM model includes two parts: model minimizing the heterogeneity between objective and sub-
Part 1: Indicator weighting phase. Based on the constructed jective weights is constructed, and the integrated weight is ob-
evaluation index system, this paper proposes a subjective and tained, so as to evaluate the 3E performance of the CCHP-MG
objective integrated weighting method based on AEW and grey- system.
DEMATEL methods. Firstly, based on the values of multiple cases Part 2: MCDM phase. The constructed indicators are all quan-
to be evaluated on all indicators, the anti-entropy value of each titative indicators and the purpose of this model is to evaluate the
indicator is calculated, and then the anti-entropy value is normal- 3E performance of multiple CCHP-MG, so this paper proposes a
ized to obtain the objective weight of the indicator. Secondly, MCDM method based on DQGRA-TOPSIS. Firstly, based on the
several experts were invited to judge the importance of each values of multiple cases to be evaluated on all indicators, a
4
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
Dmax k k
min ðkÞ ¼ max5xij min5 xij (8) (3) Integrated weight based on subjective and objective weights
i;j i;j
Step 4: Calculate the real-numbered direct influence matrix. The subjective weighting method ranks the importance of
Integrating the clarified direct grey relationship matrices Z k ¼ evaluation indicators according to the experts' own experience and
½zkij nn of all experts (k ¼ 1; 2; …; K represents the number of ex- knowledge, which reflects the subjective willingness of the experts
perts), the real-numbered direct influence matrix X ¼ ½xij nn is: on the importance of evaluation indicators, so the weights obtained
are more interpretable. However, subjective weight does not reflect
the data information of evaluation indicators. The objective weight
1 X
K
xij ¼ zkij (9) reflects the data information of the evaluation indicators but it will
K change with the change of the evaluation object set, that is, the
k¼1
stability of objective weight is weaker than that of subjective
Step 5: Determine the comprehensive influence matrix of
weight. Meanwhile, the objective weight cannot reflect the
DEMATEL. Based on the real-numbered direct influence matrix X ¼
importance of the evaluation indicator itself, so its explanatory
½xij nn , the normalized direct influence matrix A ¼ ½aij nn can be
power is weaker. Based on the respective advantages and disad-
calculated by formula (10): vantages of subjective and objective weighting methods, many
scholars have investigated the integrated weighting method, hop-
xij
aij ¼ Pn ; n ¼ 13 (10) ing that through integrated weighting, it will retain the advantages
max j¼1 xij of subjective and objective weighting method and overcome the
0in
shortcomings of subjective and objective weighting method
For matrix A, there is lim Ak ¼ O (O is zero matrix), so the [51e53]. Under this premise, using integrated weighting method to
k/∞
comprehensive influence matrix T can be calculated by: determine the index weight, on the one hand, can avoid the situ-
ation that experts’ scoring is too subjective, on the other hand, can
also avoids the situation that objective weighting ignores the index
T ¼ lim A þ A2 þ A3 þ … þ Ak ¼ AðI AÞ1 (11) attributes and causes the weight distribution to be unreasonable, so
k/∞
as to obtain a more effective weight.
where I is the unit matrix. In this paper, the fundamental principle of integrating the
Step 6: Calculate the center degree Mi and the cause degree Ni of subjective and objective weights is to minimize the heterogeneity
each indicator according to the following formula: between objective and subjective weights. Therefore, the
6
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
integrated weights are obtained by solving the following optimi- Secondly, construct the grey relation degree of geometric simi-
zation problems: larity according to Drijþ and Drij :
n h
X i .
min ðw0i w1i Þ2 ðw0i w2i Þ2 gþ sþ ; rij ¼ 1
1j i
exp Drijþ (23)
i¼1
(17)
Xn
.
s:t: w0i ¼ 1
g
1j si ; rij ¼ 1 exp Drij (24)
i¼1
where w0i is the normalized integrated weight of indicator i. In this where gþ ðsþ ; rij Þ and g
1j i
1j ðsi ; rij Þ are the grey relation degrees of
way, the total distance between the obtained integrated weight and geometric similarity of case j regard to the positive and negative
the subjective/object weights is the smallest, thereby balancing the ideal solutions, respectively.
indicator importance degrees reflected by the subjective weights Thirdly, construct the grey relation degree of numerical simi-
and the objective weights, that is, the integrated weight can not larity according to Drijþ and Drij :
0 0
only reflect the attributes of the indicator itself, but also make full
use of the indicator data information. . 0
gþ sþ ; rij ¼ 1
2j i
exp 1 Drijþ (25)
S ¼ s 0 gþ
1 ; s2 ; …; si ; …; sn ; si ¼ min rij (20) j
dj ¼ (29)
j
g
j
þ gþ
j
Step 4: Calculate the comprehensive grey relation degrees of
A larger dj value means that case j is relatively close to the
standardized case regard to the positive and negative ideal solu-
positive ideal solution and relatively far away from the negative
tions. Firstly, calculate the difference and quotient of each indicator
ideal solution. If the dj of a case is the largest, this case is the optimal
in the standardized case regard to the positive and negative ideal
case among the alternatives.
solutions [54,55]:
Drijþ ¼ sþ
i rij ; Drij ¼ si rij (21)
3.4. Applicability of the proposed hybrid MCDM model
0
. .
0
Drijþ ¼ rij sþ
i ; Drij ¼ s
i rij (22) The proposed hybrid MCDM model for evaluating the 3E per-
formance of CCHP-MG system based on AEW, grey-DEMATEL and
where Drijþ and Drij are the differences of indicator i in case j regard DAGRA-TOPSIS techniques mainly includes two modules: one is to
0 weight the indicators based on AEW and grey-DEMATEL methods,
to the positive and negative ideal solutions, respectively. Drijþ and
0
and the other is to rank the evaluated cases employing the DQGRA-
Drij are the quotients of indicator i in case j regard to the positive TOPSIS technique. The applicability of the proposed hybrid model is
and negative ideal solutions, respectively. mainly reflected in:
7
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
(1) Applicability and advantages of the index weighting on the differences of evaluation objects has good applicability [62].
methods Secondly, the conventional TOPSIS makes Euclidean distance as
the closeness between the alternative to be evaluated and the
Firstly, considering that all indicators in the constructed index positive and negative ideal solutions. When two alternatives are
system is quantitative, and this paper takes a number of typical located on the vertical bisector of the positive and negative ideal
CCHP-MG as the cases, the objective weighting method with solutions (Fig. 5), their Euclidean distances for the positive ideal
difference-driven can be adopted for indicator weighting. Entropy solution is equal to the Euclidean distances for the negative ideal
weight method is a widely used objective weighting method, solution, which makes it impossible to rank them [63]. In this case,
assuming that the difference of indicators is inversely proportional introducing the GRA method into TOPSIS can effectively overcome
to the entropy value and proportional to the weight coefficient of this defect [64,65].
the indicator, so the weights of the indicators can be determined by Finally, the basic idea of GRA is to judge the degree of relevance
calculating the entropy value of the MCDM system [47]. However, based on the similarity of the data sequence, which is a method
in the conventional entropy weight method, the index difference proposed by the grey system theory to analyze the relevance de-
reflected by the entropy value is too sensitive, which may lead to gree of various factors in the system. However, the determination of
the extreme situation of the index weight being too small in the the resolution coefficient in conventional GRA method is subjec-
weight distribution. The AEW method is different from the entropy tive. Furthermore, conventional GRA uses the absolute value of the
method, and the difference of the indicators is proportional to the difference between two data sequence to calculate the relation
calculated anti-entropy value and the weight coefficient, which can degree, only considering the degree of geometric similarity be-
avoid the emergence of extreme situations [48,56]. Therefore, this tween the data sequence, and ignoring the numerical closeness
paper employs the AEW method to determine the objective degree [54,55]. If the two curves are parallel, the grey relation de-
weights. On one hand, it makes full use of the original information gree calculated by conventional GRA is 1. In fact, the two curves are
of the indicators. On the other hand, it avoids the extreme situation not completely consistent, so the calculated grey relation degree is
of indicator weights being too small in the weight distribution. biased. Therefore, this paper proposes a GRA method that combines
Secondly, as an objective weighting method, the AEW method difference and division methods, defining the comprehensive grey
pays too much attention to the difference between indicators while relation degree from geometric and numerical similarity, which
ignores the meaning of the indicator itself. Therefore, it is necessary overcomes the limitation of conventional grey relation degree that
to modify the objective weight through the subjective weighting only considers geometric similarity. Besides, the improved GRA
method [57]. Economy, environment and energy are an interactive does not need to set the resolution coefficient, avoiding the
organic system. In the 3E performance evaluation index system subjectivity in the calculation process.
constructed in this paper, there is a certain correlation between
each indicator, and the importance of the indicator mainly depends 4. Case analysis
on the status of the indicator in the 3E system, that is, the inter-
action relationships with other indicators. Therefore, the DEMATEL 4.1. Basic information of the cases
technique, as an effective method that can reflect the interactions of
factors in complex systems and determine the status of factors [58], In this section, 8 building-typed CCHP-MG systems located in a
is suitable for the subjective weighting of this paper. However, the demonstrative industrial park of Jiangsu Province are employed for
conventional DEMATEL method inevitably faces problems such as case analysis, and the composition of each CCHP-MG system is
insufficient information and uncertainty when judging the corre- shown in Table 2. According to Table 2, the typical characteristics of
lation between factors, making the judgment result more subjec- each case are calculated (Fig. 6), from which it can be seen that: (1)
tive [59]. In this case, this paper introduces the grey system theory Cases 2 and Case 3 have the largest distributed renewable energy
to improve the conventional DEMATEL method, which can deal installed capacity of 100 kW, while Case 7 has the smallest installed
with the uncertainty and ambiguity in the process of factor judg- capacity of 40 kW. However, from the perspective of the proportion
ment, so as to improve the accuracy of factor judgment [50,59,60]. of distributed renewable energy in the total energy supply, Case 4 is
Finally, according to Li G et al. [61], the criteria for judging the the largest (13.79%) and Case 1 is the smallest (6.67%). (2) Case 3 has
rationality of subjective-objective combination weights include: the largest total energy storage capacity, reaching 600 kW, while
the combination mode of subjective and objective weights is Case 7 has the smallest total energy storage capacity (180 kW). (3)
reasonable, the interpretability of the combination weights should Case 3 has the largest total energy demand and supply, reaching
be ensured, and the evaluation effect is better. Therefore, on the 780 kW and 950 kW, respectively, while Case 7 has the smallest
basis of the AEW weights and the grey-DEMATEL weights, this total energy demand and supply of 290 kW and 360 kW,
paper makes normalizing the integration weights be the constraint,
and minimizing the heterogeneity of the subjective and objective
weight results be the objective function, to establish an optimiza-
tion model, and solves it to obtain the integrated weight, which
makes full use of the original information of the indicators, and at
the same time considers the correlation between the indicators on
the basis of reducing subjectivity, ensuring the reliability of the
weight results. Overall, the combination mode of the integrated
weight is reasonable, and the integrated weight can be well
explained.
Table 2
Composition of 8 building-typed CCHP-MG systems.
respectively. (4) Case 2 has the largest system energy supply- were invited to issue questionnaires to obtain each expert's judg-
demand ratio of 1.4545, and Case 8 has the smallest one of 1.1522. ment on the correlation between indicators. Then, the direct grey
relationship matrix and the clarified judgment matrix of each
4.2. Results of indicator weighting expert can be calculated by formulas (4)~(7). On this basis, ac-
cording to formulas (9)~(11), the comprehensive influence matrix T
(1) Determination of objective weight based on AEW method of grey-DEMATEL is calculated, as shown in Table 4.
Based on the comprehensive influence matrix, according to
For the 8 building-typed CCHP-MG systems, collect the indicator formulas (12)~(16), the subjective weight based on grey-DEMATEL
values of each case, and then calculate the objective weight of each is calculated, and the results are shown in Table 5.
index according to formulas (2) and (3), as shown in Table 3. On the basis of determining the weights based on AEW and
In this paper, 9 experts (3 from power grid enterprises, 4 from grey-DEMATEL, using MATLAB programming to solve the formula
universities, and 2 from government energy regulatory agencies) (17), the integrated weight are obtained:
Table 3
Calculation result of objective weight based on AEW method.
Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8 hi w1i Ranking
c11 13.31 17.35 19.32 21.07 16.8 14.12 14.44 9.72 0.1406 0.0743 11
c12 17.15 15.01 12.69 10.44 15.87 25.69 8.35 17.85 0.1497 0.0791 4
c13 16.37 18.83 20.45 10.98 21.37 7.73 13.44 12.25 0.1473 0.0778 6
c14 40.64 47.7 61.11 35.54 33.49 52.89 22.78 37.91 0.1449 0.0765 9
c15 167.5 189.81 215.73 120.16 157.01 141.84 72.04 133.4 0.1451 0.0767 8
c21 153.29 177.11 229.08 135.92 120.36 139.1 103.32 132.59 0.1429 0.0755 10
c22 0.065 0.083 0.098 0.043 0.078 0.055 0.033 0.063 0.1481 0.0783 5
c23 0.34 0.289 0.506 0.232 0.273 0.306 0.123 0.223 0.1538 0.0813 2
c24 0.121 0.13 0.192 0.115 0.129 0.099 0.043 0.109 0.1500 0.0793 3
c31 4.78 7.64 8.55 12.88 6.04 5.13 5.04 5.22 0.1559 0.0824 1
c32 87.96 91.61 89.1 93.09 82.9 89.64 92.34 86.02 0.1337 0.0707 13
c33 562.63 600.33 623.83 541.18 585.37 528.69 635.37 612.42 0.1341 0.0709 12
c34 386.32 320.01 443.88 286.75 364.3 261.98 160.44 210.66 0.1463 0.0773 7
Note: The values of each indicator in the 8 cases were collected manually by the authors.
(2) Determination of subjective weight based on grey-DEMATEL
9
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
Table 4
The comprehensive influence matrix of grey-DEMATEL.
c11 c12 c13 c14 c15 c21 c22 c23 c24 c31 c32 c33 c34
c11 0.0221 0.0892 0.0696 0.0067 0.0179 0.0155 0.0136 0.0087 0.0134 0.0403 0.0122 0.0045 0.0039
c12 0.0977 0.0212 0.0785 0.0045 0.0333 0.0372 0.0273 0.0229 0.0272 0.0160 0.0140 0.0024 0.0033
c13 0.0904 0.0737 0.0224 0.0039 0.0193 0.0135 0.0074 0.0067 0.0072 0.0164 0.0148 0.0031 0.0084
c14 0.1235 0.0619 0.0581 0.0092 0.0943 0.0363 0.0321 0.0313 0.0317 0.0077 0.0262 0.0226 0.0591
c15 0.1287 0.1115 0.1093 0.0578 0.0146 0.0170 0.0154 0.0144 0.0115 0.0076 0.0071 0.0039 0.0187
c21 0.0457 0.0444 0.0350 0.0027 0.0342 0.0092 0.0555 0.0551 0.0555 0.0027 0.0072 0.0009 0.0018
c22 0.0473 0.0456 0.0361 0.0036 0.0552 0.0469 0.0112 0.0668 0.0635 0.0027 0.0027 0.0006 0.0016
c23 0.0468 0.0452 0.0357 0.0033 0.0513 0.0469 0.0671 0.0108 0.0635 0.0027 0.0026 0.0006 0.0015
c24 0.0361 0.0348 0.0293 0.0029 0.0457 0.0415 0.0542 0.0539 0.0094 0.0021 0.0023 0.0005 0.0013
c31 0.1858 0.1435 0.1322 0.1230 0.1699 0.1413 0.1345 0.1331 0.1297 0.0125 0.1039 0.0755 0.0573
c32 0.1881 0.1781 0.1709 0.0760 0.1575 0.0975 0.0925 0.0910 0.0878 0.0285 0.0229 0.0900 0.1295
c33 0.1028 0.0991 0.0883 0.0151 0.0851 0.0610 0.0478 0.0470 0.0435 0.0085 0.0532 0.0083 0.0340
c34 0.1421 0.1391 0.1167 0.0399 0.1400 0.1552 0.1570 0.1558 0.1562 0.0201 0.0856 0.1182 0.0182
10
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
Table 6
The weighted decision matrix of TOPSIS.
Indicator number Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8
11
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
of the system. Moreover, it is much needed to increase the scale of to provide a reference for improving the 3E benefits of CCHP-MG
renewable energy installed capacity and the penetration rate of system.
renewable energy in the system, which can enhance the energy As suggested by Bhatia and Srivastava [66] and Zhang and Deng
utilization efficiency of the system and improve the benefits of [67], taking the center degree Mi and cause degree Ni of each
energy saving and emission reduction.
5. Discussions
Table 7
The models for ranking similarity comparison.
Model 1 The proposed hybrid MCDM model based on AEW-grey-DEMATEL and DQGRA-TOPSIS
Model 2 The hybrid MCDM model based on AEW and DQGRA-TOPSIS
Model 3 The hybrid MCDM model based on grey-DEMATEL and DQGRA-TOPSIS
Model 4 The hybrid MCDM model based on AEW-grey-DEMATEL and conventional TOPSIS
Model 5 The hybrid MCDM model based on AEW-grey-DEMATEL and conventional GRA-TOPSIS
12
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
indicator in Table 7 as the X-axis and Y-axis, the causal relationship The similarity of the rankings is an important indicator to reflect
of each indicator in the 3E system of the CCHP-MG is drawn the robustness of the proposed MCDM method, which can usually
(Fig. 10). be measured by the similarity coefficient of rankings [68]. Ac-
According to Fig. 10, several findings are discovered: cording to Sałabun et al. [69] and Kizielewicz et al. [70], there are
three commonly used ranking similarity coefficients, named
(1) The indicators in the 3E comprehensive performance system Spearman's rank correlation coefficient (rS ), weighted Spearman's
of the CCHP-MG can be divided into four areas according to rank correlation coefficient (rW ), and WS similarity coefficient
the center and cause degrees, that is, L-H area, HeH area, L-L (WS), which can be calculated by formulas (30)-(32), respectively.
area and H-L area. Among them, the indicators being in the
PN
HeH area can be regarded as the causal factors with greater 6 ðxi yi Þ2
importance in the system, the indicators being in the L-H rS ¼ 1
i¼1 (30)
N N2 1
area can be regarded as the causal factors while with little
importance in the system, the indicators being in the H-L
area can be regarded as important result factors in the sys- PN
6 i¼1 ðxi yi Þ2 ½ðN xi þ 1Þ þ ðN yi þ 1Þ
tem, and the indicators being in the L-L area can be regarded rW ¼ 1 (31)
as the result factors with little importance in the system. N N3 þ N2 N 1
(2) The three indicators named equivalent saved energy con-
N
sumption, system energy cascade utilization rate and
X jxi yi j
renewable energy penetration rate are in the HeH region, all WS ¼ 1 2xi (32)
of which are energy performance indicators, indicating that i¼1
maxfjxi 1j; jxi Njg
the energy utilization level and renewable energy scale are
the key factors in the 3E comprehensive performance system where xi and yi are the sequences of rankings, and N is the number
of the CCHP-MG system, which can affect the 3E perfor- of alternatives in the rankings.
mance of the system by acting on other factors. The three It is worth noting that the Spearman's rank correlation coeffi-
indicators named return rate of power supply, return rate of cient and the weighted Spearman's rank correlation coefficient are
heating supply and investment saving benefits are in the H-L both symmetrical coefficients, which means that the position of the
region, so they are the key result factors of the system's 3E difference between the two rankings has no significant impact on
comprehensive performance, indicating that energy supply the similarity of the rankings. For instance, if in two ranking se-
benefits and investment savings benefits are the most intu- quences, the rankings of only three alternatives are different, and
itive benefits of the CCHP-MG system. These indicators are the ranking results of the other alternatives are the same, then
affected by other factors, and ultimately affect the 3E whether the three alternatives are the top three or the bottom three
comprehensive performance of the system. All in all, it is will have no effect on the similarity of the two ranking sequences.
necessary to focus on the indicators in HeH and H-L areas, However, in the actual MCDM issues, whether the ranking of the
which is of great value for improving the 3E comprehensive top alternatives is consistent is more important for measuring the
performance of the CCHP-MG system. robustness of the MCDM results. The WS similarity coefficient, as an
(3) The center degrees of indicators in the L-H and L-L regions asymmetrical similarity coefficient, holds that the ranking differ-
are small, indicating that the importance of these indicators ences of the top-ranked alternatives have more influence on the
in the system is relatively weak. It is difficult for each one of final ranking similarity.
these factors to independently affect the 3E comprehensive This paper sets up 4 contrastive hybrid MCDM models for
performance of the CCHP-MG system. Therefore, these fac- ranking similarity comparison, as shown in Table 7. The steps of
tors need to be more or less related to other factors in order conventional TOPSIS and conventional GRA-TOPSIS can be referred
to pass on their own influence on the 3E performance of to Behzadian et al. [62] and Baranitharan et al. [64] respectively. By
CCHP-MG system or help other factors pass on the influence. comparing the ranking similarity between Model 1, 2 and 3, the
All the environmental performance indicators are in the L-L robustness of the proposed weighting methods can be analyzed,
region, indicating that they are all result factors. These in- and by comparing the ranking similarity between Model 1, 4 and 5,
dicators are affected by energy utilization level, equivalent the robustness of the designed MCDM phases can be analyzed.
energy saving and other indicators, thus having an impact on Based on the relevant data of the 8 building-typed CCHP-MG
the 3E comprehensive performance of the CCHP-MG system. systems, the ranking results under different models in the overall
performance dimension and the 3E performance dimension are
calculated. Then, according to formulas (30)~(32), the similarity
coefficients between the ranking results of proposed MCDM model
5.2. Effectiveness check of rankings and each contrastive model are calculated, as shown in Table 8.
According to Table 8, it can be found that:
Verifying the effectiveness of the MCDM model from an
empirical perspective mainly includes two dimensions: First, 1) In terms of overall performance, the rankings of Model 1 and
compared with other models, the ranking results of the proposed the 4 contrastive models are highly consistent. After chang-
model should not be significantly different from the results of other ing the weighting method and the MCDM phase, the overall
models, unless it can be proved theoretically and practically that performance ranking results have not changed, indicating
the results of other models are significantly biased, that is, the that the proposed MCDM model has better robustness when
proposed model should have better robustness. Second, compared evaluating the 3E performance of the examined building-
with other models, the proposed model ranking results should have based CCHP-MG systems.
greater differences between alternatives, that is, the proposed 2) The similarity coefficients of Model 2 and 3 show that for
model has better sample discrimination. economic and environmental performance, the ranking re-
sults of Model 2 and 3 relative to Model 1 are highly
(1) Robustness check of rankings consistent. In energy performance dimension, the ranking
13
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
Table 8
The similarity coefficients between the rankings of the proposed model and the contrastive models.
Table 9
The sample discrimination test indicators of the proposed model and the contrastive models.
Standard deviation (s) Relative range (q) Coefficient of variation (w) Sensitivity (h)
15
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
16
H. Zhao, B. Li, H. Lu et al. Energy 240 (2022) 122830
[66] Bhatia MS, Srivastava RK. Analysis of external barriers to remanufacturing improved TOPSIS gray correlation projection method. Electr Power Auto
using grey-DEMATEL approach: an Indian perspective. Resour Conserv Recycl Equip 2019;39(4):63e9.
2018;136:79e87. [72] Aytekin A. Comparative analysis of the normalization techniques in the
[67] Zhang W, Deng Y. Combining conflicting evidence using the DEMATEL context of MCDM problems. Decision Making: Appl Manag Eng 2021;4(2):
method. Soft Comput 2019;23(17):8207e16. 1e25.
[68] Sałabun W, Urbaniak K. A new coefficient of rankings similarity in decision- [73] Vavrek R. Evaluation of the impact of selected weighting methods on the
making problems. In: International conference on computational science. results of the TOPSIS technique. Int J Inf Technol Decis Making 2019;18(6):
Cham, Switzerland: Springer; 2020. p. 632e45. 1821e43.
[69] Sałabun W, Wa˛ trobski J, Shekhovtsov A. Are MCDA methods benchmarkable? [74] Ge L, Li Y, Li S, Zhu J, Yan J. Evaluation of the situational awareness effects for
a comparative study of TOPSIS, VIKOR, COPRAS, and PROMETHEE II methods. smart distribution networks under the novel design of indicator framework
Symmetry 2020;12(9):1549. and hybrid weighting method. Front Energy 2020:1e16.
bski J, Sałabun W. Identification of relevant criteria set in
[70] Kizielewicz B, Wa˛ tro [75] Dutta B, Singha T, Goh M, Lamata MT, Verdegay JL. Post factum analysis in
the MCDA processdwind farm location case study. Energies 2020;13(24): TOPSIS based decision making method. Expert Syst Appl 2019;138:112806.
6548. [76] Zhang K, Zhan J, Wang X. TOPSIS-WAA method based on a covering-based
[71] Liang H, Tian S, Li Q, Liu Y, Zhang W. Main grid structure evaluation based on fuzzy rough set: an application to rating problem. Inf Sci 2020;539:397e421.
17