Shu－Ming Ma，Shu－Shen Zhang，Yu Chen，Guo－Bao Song，Peng Gao

(1.Faculty of Chemical，Environmental and Biological Science and Technology，Dalian University of Technology，Dalian 116024，China; 2.Center for Post-doctoral Studies of Civil Engineering，Harbin Institute of Technology，Harbin 150001，China)

1 Introduction

Chemical Industry Park(CIP)has evolved into the mainstream of chemical industry both domestic and international for its advantage in strong correlation between products of the plants within，sharing the public facilities and centralized pollution treatment［1］.However，the risk of major accidents of CIP is huge.In a CIP，chemical enterprises are concentrated and large amount of flammable，explosive and toxic chemicals are stored.When a major accident occurs，a prompt and effective emergency response is crucial for preventing or reducing negative consequences.Thus，to strengthen the emergency response capacity is a very important safety precaution issue for CIPs.Based on this fact，this paper focuses on studying on the method of evaluating emergency response capacity of Chemical Industrial Park(ERCCIP)in order to provide decision makers of CIPs a scientific tool to measure and improve the level of ERCCIP.

In the last decade，the security issues of China’s CIP has become major concern of the society after several serious accidents［2］.Some scholars have carried on the research on the ERCCIP.Zeng and Wu studied emergency management mode of CIP［3］.Ji and Yang carried out evaluation on emergency resource supporting capability of CIP by a fuzzy AHP model［4］.ERCCIP based on the AHP method from different perspectives were studied in Refs.［5－7］.It is seen form literatures that ERCCIP has become a hot research field.For ERCCIP is essentially a multi-criteria decision problem and involves expert’s subjective judgment，fuzzy AHP is frequently used［8］.

Fuzzy AHP，as a fuzzy extension of AHP，was proposed by van Laarhoven and Pedrycz and extended by other authors［9］.Despite its popularity and merit in dealing with decision makers’ ambiguities，uncertainties and vagueness，there are criticisms about those works for they are unable to deal with the issue of consistency［10］.With this problem at hand，a fuzzy preference programming(FPP)wasproposedin Mikhailov’s research［11］and improved by Rezaei et al.［9］.This FPP method not only guarantees the preservation of the preference intensities，but also provides a well interpretive consistency index.Given its tested capability in tackling the issue of consistency，our study employs it to determine the weights of the fuzzy AHP model for ERCCIP.

Besides the improved FPP，the 2-tuple linguistic representation model［12］was also introduced in the fuzzy AHP model for ERCCIP.Because during the evaluation process，there are situations in which experts could not provide numerical precise information due to time constrains， inadequate of data or vague knowledge.The linguistic representation modelis presented by Herrera and Martínez based on the concept of symbolic translation［12］and has been widely used during the lastdecade for its accuracy，interpretability，usefulnessforimproving linguistic solving processes， ease managementofcomplex frameworks in which linguistic information is included.

The main purpose of this paper is to propose a method to evaluate ERCCIP.The method consists of three main steps:

1)developing evaluation index system of ERCCIP;

2)determining the weights of evaluation index system using the improved FPP;

3)Computing the aggregated ratings of criteria and the overall ERCCIP.

2 Evaluation Index System of ERCCIP

Evaluation index system of ERCCIP is shown in Fig.1.It is constructed based on related literatures and national guidelines for Environmental major accidents［4－8］.The overall goal is to evaluate ERCCIP as shown in first level.The second level is the criteria to fulfill the overall goal.The general criteria consists of fourmajorcriteria which are“early warning capability”(B1)，“emergency process capability” (B2)，“after-accidents process capability”(B3)and“emergency support capability”(B4).These criteria are further decomposed into specific items in the third level.The“earlywarningcapability” (B1)is decomposed into five sub-criteria: “information monitoring capacity” (C11)， “major hazard installations identification capacity”(C12)，“accident scenarios identification capacity”(C13)，“emergency plans”(C14)and“early warning support system”(C15).The sub-criteria for“emergency process capability” (B2) are “emergency response mechanism”(C21)，“emergency monitoring”(C22)，“information dissemination capability” (C23)，“emergency rescue”(C24)，and“accident source disposal capability” (C25).The“after-accident process capability”(B3)is decomposed into three sub-criteria:“evaluationandsummary”(C31)，“reconstruction capability”(C32)and“insurance”(C33).The“emergency support capability”(B4) consists offive sub-criteria:“financialsupport”(C41)，“equipment support”(C42)，“communication support”(C43)，“human resources support”(C44) and“technical support”(C45).

Fig.1 Evaluation index system of ERCCIP

3 Determining the Weights of Evaluation Index System Using the Improved FPP

The improved FPP［9］is employed to determine the weights of criteria and sub-criteria of the evaluation index system of ERCCIP.First，pairwise comparison matrices are constructed by comparing the importance of criteria or sub-criteria.A typical pairwise comparison matrix is as follows:

where~aijis a triangular fuzzy number(TFN)shown in Fig.2 which indicates the experts’preference of i over j andThe definition of TFN is given in Definition 1.

Fig.2 Triangular fuzzy numbers(TFNs)

Definition 1［13］We define a fuzzy number N on R(=(－∞，∞))to be a triangular fuzzy number if its member function μN(yùn):R→［0，1］is equal to:

where l≤m≤u，and m stands for the modal value，l and u are the lower and upper values of the support of N respectively.The triangular number，as given by Eq.(2)，will be denoted by(l，m，u).

After the pairwise comparison matrices are constructed，the priority vectors ω=(ω1，ω2，…，ωn)Tcan be obtained by solving an improved FPP.ω1，ω2，…，ωnare the crisp value of weights for the comparison matrices.The theory of the FPP was detailed described by Rezaei et al.［9］.Here we just list the equations as follows:

max λ

s.t.

Fuzzy numbers of type I refers toand fuzzy numbers of type II refers toBy solving the above equation the optimal priority vector ω*and λ*are obtained.λ*can be interpreted as consistency index.The negative value of λ*indicates that the pairwise comparisons are strongly inconsistent while the positive value of λ*indicates that the pairwise comparisons are consistent.If λ*=1，it can be concluded that the pairwise comparisons are fully consistent［9］.

4 Computing the Aggregated Ratings of Criteria and the Overall ERCCIP

4.1 2-Tuple Linguistic Representation Model

The linguistic representation model［12］is by using 2-tuple(si，α)，where siis on the elements of set S= {s0，s1，s2，…，sg}，and α is a numerical value representing the distance to the central value of the i－th linguistic term.

Definition 2［12］Let S={s0，s1，s2，…，sg}be a linguistic term set，and β∈［0，g］is a value representing the resultofa symbolic aggregation operation，then the 2-tuple linguistic information equivalent to β is obtained with the following function:

with

where“round”stands for the usual round operation;sihas the closest index label to“β”，and“α”is the value of the symbolic translation.

On the contrary，the 2-tuple linguistic variable can be converted into an equivalent numerical value β(β∈［0，g］)by the following formula

where Δ－1stands for a reverse equation for converting the 2-tuple linguistic variable into a crisp value β.

In this study，a five terms S={s0，s1，s2，s3，s4} is defined for assessing ERCCIP.The meanings of linguistic terms s0，s1，s2，s3，s4are“Very Low”，“Low”，“Average”，“High”，and“Very High”.

4.2 Assessing the Ratings of Sub-Criteria

After the weights are determined，the experts evaluate the ratingsofsub-criteria by using the linguistic items(“Very Low”，“Low”，“Average”，“High”，and“Very High”).After that，the linguistic items are transformed into 2-tuples using Eq.(8).Then the rating of sub-criteria can be calculated based on arithmetic mean operator as follows.

4.3 Computing the Aggregated Ratings of Each Criteria

The aggregated ratings of criteria are computed by the following equation.

where(si，αi)is 2-tuple for aggregated rating of i－th criterionis one element of the set S;αi∈［－0.5，0.5);niis the number of sub-criteria with respect to i－th criterion.

4.4 Computing the Overall ERCCIP

The overall ERCCIP could is computed by the following equation.

where(s0，α0)is 2-tuple for aggregated rating of overall ERCCIP，βi=Δ－1(si，ai)，s0is on the elements of the set S;α0∈［－0.5，0.5);n is the number of criteria.

5 Case Study:Evaluating the Emergency Response Capacity of Songmudao Chemical Industrial Park

In this section，we apply the proposed method to Songmudao Chemical Park which is located in the Midwest of Dalian City and Northeast of China as shown in Fig.3.It has five core industry activities: marine chemicals， specialty petrochemicals， fine chemicals，synthetic gas chemical and post processing.All the five industry activities involve large numbers of dangerous chemical installations.Take synthetic gas chemical for example，main productions of this activity are ammonia，methanol，etc.The consequences will be very serious once accidents occurred in those chemical installations without effective emergency response.

This case study is carried on the evaluation on the level of ERCCIP of Songmudao Chemical Industrial Park to provide recommendations for improving safety precautions of the park，meanwhile，serve as an illustration of the method.

Fig.3 Location of Songmudao Chemical Park

The evaluation index system for ERCCIP shown in Fig.1 is employed here and a decision-making group consisting of seven experts(E1，E2，…，E7)is formed.The first task is to determine the weights of the evaluation index system.A joint meeting is hold to make comparisons and construct fuzzy comparison matrices by using triangular fuzzy numbers.First of all，the criteria and sub-criteria of ERCCIP are explained to the team in great details，in order that the team can provide objective and precise responses.Pairwise comparison matrices are formed by the expert team using the scale given in Table 1.Fuzzy evaluations are performed in the pairwise comparisons by the expert team as follows:“early warning capability”(B1)and“emergency process capability”(B2)are compared using the question‘‘How important are early warning capability(B1)when it is compared to emergency processcapability (B2)” and ifthe answeris“Strongly more important”，for this linguistic scale the triangular fuzzy number is(4，5，6).All the fuzzy evaluation matrices are produced in the same manner.The comparison results are shown in Tables 2 and 3.

Table 1 Linguistic scales for importance

Table 2 Pairwise comparison matrix and weights of criteria

Table 3 Local weights and pairwise comparison matrix of sub-criteria

As depicted in Section 3，crisp weights for criteria and sub-criteria are obtained through solving the improved non-linear programming model as shown in Eq.(6).Consistency index λ*for each matrix are as follows:λA=0.838，λB1=0.525，λB2=0.442，λB3=0.643，λB4=0.442，whereandare consistency indexes of matrix for criteria and four matrixes for sub-criteria respectively.All the consistency indexes are within the range of 0－1 which means the initial judgments are consistent.

The weighs of criteria are listed in column 6 of Table 2.Based on the experts’judgment，“Emergency process capability(B2)”is considered the most important dimension among the four criteria.The weights of sub-criteria are listed in column 7 of Table 3.For“early warning capability(B1)”dimension，“emergency plans(C14)”is ranked as the most important item，for“emergency process capability (B2)”dimension，“emergencyrescuecapability (C24)”is ranked as the most important item，for“after-accident process capability(B3)”dimension，“reconstruction capability(C32)”is ranked as the most important item，and finally for the“emergency supportcapabilities(B4)” dimension，“human resources support(C44)”is ranked as the most important item.The local weights of sub-criteria are obtained by multiplication of the weight of sub-criterion and the weight of the main criterion to which it belongs and listed in column 8 of Table 3.It is seen that items“emergency rescue capability(C24)”is ranked as the most important item.The conclusions are consistent with the results calculated by conventional AHP as shown in column 9 of Table 3.The values of the fuzzy AHP model proposed are quite close to the values calculated using traditional AHP.This situation shows that the proposed model provides valid results as the traditional AHP does.On the other hand，this proposed method using fuzzy theory is more suitable to deal with decision makers’ ambiguities， uncertainties and vagueness comparing conventional AHP.

In the next stage，each expert evaluated the ratings of sub-criteria by linguistic items(“Very Low”，“Low”，“Average”，“High”，and“Very High”).The results are given in Table 4.The 2-tuples for the ratings of the sub-criteria are calculated based on Eq.(6)and shown in the column 9 of Table 4.

Table 4 Sub-criteria rating by experts

The aggregated ratings of each criteria and the overall ERCCIP are calculated by Eqs.(7)and(8).The results are shown in Table 5.The level of four criteria of Songmudao Chemical Park can be interpreted as higher than“average”，higher than“average”，higher than“average”and lower than“average”respectively.The obtained overall ERCCIP of Songmudao Chemical Park can be interpreted as higher than “average”.“Emergency support capability”(B4)should be ranked as priorities for Songmudao Chemical Park to improve its ERCCIP in the future.Based on the index emergency support capability can be strengthened from five apsects:“financial support”(C41)，“equipment support”(C42)，“communication support”(C43)，“human resources support”(C44)and“technical support”(C45).Indicated by the results shown in Table 4，strengthening human resources support capability should be ranked as priorities for improving the level of emergency support capability of Songmudao Chemical Park.The following measures should be taken:(i)the establishmentofa professionalemergency rescue teams;(ii)strengthening skills training of rescue teams.

Table 5 Result of overall emergency response capacity

6 Conclusions

In this paper，we present a fuzzy AHP method by combining an improved fuzzy preference programming and 2-tuple fuzzy linguistic to evaluate the ERCCIP.The main features of the method are summarized as follows:

1)A hierarchy of the evaluation index system for ERCCIP is presented.

2)An improved FPP is employed to determine the weights of the criteria and sub-criteria.

3)2-Tuple linguistic representation model is used toexpressthe experts’ subjective opinionsand compute aggregated ratings of criteria and the overall ERCCIP.This hybrid fuzzy AHP method has the ability to deal with the issue of consistency of conventional fuzzy AHP and capture the vagueness of human thinking style.Particularly，it provides decision makers a scientific tool to measure and improve their ERCCIP or select the best alternative.

［1］Chen Y.Study on the Environmental Risk Assessment Method and Risk Management of Chemical Industrial Park Planning.Dalian:Dalian University of Technology，2013.

［2］Ma S M，Zhang S S，Chen Y，et al.Design and realization of a major environmental risk source management system.Procedia Environmental Sciences，2013，18:372－376.

［3］Zeng M R，W Z Z，Wei L J，et al.Study on emergency management model for accidents in chemical industry park.China Safety Science Journal，2009，19(2):172－176.

［4］Ji S W，Yang X.Study on emergency capability evaluation of public emergency resources for chemical industrial park.Liaoning Chemical Industry，2013，42(8):1003－1006.

［5］Yuan F，Wu Y W，Liu J，et al.Research on emergency capability assessment of chemical industrial park based on improved AHP.Journal of Safety Science and Technology，2009，5(1):160－164.

［6］ Tan P C.Research on Comprehensive Evaluation of Accidents Emergency Capability of Chemical Industrial Park.Chongqing:Chongqing University，2009.

［7］Yu D L.Research on Evaluation of Accidents Emergency Capability and Management of Chemical Industrial Park.Dalian:Dalian University of Technology，2010.

［8］Ju Y B，Wang A H，Liu X Y.Evaluating emergency response capacity by fuzzy AHP and 2-tuple fuzzy linguistic approach.Expert Systems with Applications，2012，39 (8):6972－6981.

［9］ Rezaei J，Ortt R，Scholten V.An improved fuzzy preference programming to evaluate entrepreneurship orientation.Applied Soft Computing，2013，13(5): 2749－2758.

［10］Leung L C，Cao D.On consistency and ranking of alternatives in fuzzy AHP.European Journal of Operational Research，2000，124(1):102－113.

［11］Mikhailov L.Deriving priorities from fuzzy pairwise comparison judgments.Fuzzy Sets and Systems，2003，134 (3):365－385.

［12］HerreraF，MartínezL.A 2-tuplefuzzylinguistic representation model for computing with words.IEEE Transactions on Fuzzy Systems，2000，8(6):746－752.

［13］Laarhoven P J M V，Pedrycz W.A fuzzy extension of Saaty’s priority theory.Fuzzy Sets and Systems，1983，11 (1):199－227.