Application of a New Hybrid Fuzzy AHP Model to the Location Choice

Abstract

The purpose of this paper is to propose a new hybrid fuzzy Analytic Hierarchy Process (AHP) algorithm to deal with the decision-making problems in an uncertain and multiple-criteria environment. In this study, the proposed hybrid fuzzy AHP model is applied to the location choices of international distribution centers in international ports from the view of multiple-nation corporations. The results show that the proposed new hybrid fuzzy AHP model is an appropriate tool to solve the decision-making problems in an uncertain and multiple-criteria environment.

1. Introduction

Modern logistics service provided by the logistics provider emphasizes quick response to customer demand. However, it is difficult to fit totally customer demands in a logistics system. In particular, in a global logistics system a lot of uncertainties and complexities exist. A global logistics system includes two important roles. One is the logistics service provider, for example, shipping carriers, international ports, and international distribution centers. Another is the logistics service demander such as multinational corporations (MNCs). The international distribution center within the international port is also one important part of a global logistics system. The shipping companies and the multiple-national corporation prefer to use high-efficiency and high-service quality international logistics centers within an international port. Therefore, it is an important and complex decision-making problem for shipping companies and multinational corporations to select a high-efficiency and high-service quality international logistics center within an international port.

Due to a shift in the global center of manufacturing to Asia since 1980s, major international ports in the Asian region have been expanded rapidly. Thus, the shipping companies and the multinational corporations focus on the location choice of international distribution centers in Asia. The major international ports in Asia include the ports of Shanghai, Singapore, Hong Kong, Shenzhen, Busan, Ningbo, Qingdao, Guangzhou, Tianjin, and Kaohsiung. In the future, the demand for cargoes in Asia will further increase. In Table 1, the container throughputs in 2010 for the world’s top 20 container ports including the ports of Shanghai, Singapore, Hong Kong, Shenzhen, Busan, Ningbo, Guangzhou, Qingdao, Dubai, Rotterdam, Tianjin, Kaohsiung, Antwerp, Klang, Hamburg, Los Angeles, Tanjung Pelepas, Long Beach, Xiamen, and Laem Chabang are 29,069, 28,430, 23,611, 22,509, 14,180, 13,146, 12,545, 12,012, 11,613, 11,145, 10,086, 9,181, 8,483, 8,146, 7,900, 7,831, 6,603, 6,263, 5,824, and 5,640 thousand TEUs, respectively. The rankings and the volumes for these major international ports from 2006 to 2010 are also shown in Table 1.

Table 1 

The container throughputs and the rankings for the world’s top 20 container ports (unit: thousand TEUs).

Many international distribution centers at major Asian ports have been established in the recent years, such as Waigaoqiao Bond Logistics Park in the port of Shanghai, Hong Kong International Distribution Center in the port of Hong Kong, Kaohsiung Yes Logistics Zone in the port of Kaohsiung, Schwartz Logistics Hub in the port of Shenzhen, Busan Logistics Park in the port of Busan, and Keppel Distripark in the port of Singapore. Therefore, it is very important for the shipping companies and the multinational corporations to evaluate the environment among these major international logistics centers in different nations, in order to design and implement an appropriate global logistics system. It is also a complex multiple-criteria decision-making (MCDM) problem under uncertain environment. AHP is an appropriate approach to solve complex multiple-criteria decision-making problem. Fuzzy sets theory method has been widely applied to the uncertain decision-making problem in the real world. Thus, this paper combines AHP and fuzzy sets theory and then proposes a new hybrid fuzzy AHP model for the location choice of international logistics centers within the international ports from the perspective of multi-nation corporations.

2. Literature Review

2.1. Environmental Evaluation Approaches

Environment evaluation approaches including the resource-based view (RBV), traditional strength-weakness-opportunity-threat (SWOT), and quantitative SWOT such as the external factor evaluation matrix (EFE), internal factor evaluation matrix (IFE), and competitive profile matrix (CPM) have been widely used. The traditional SWOT analytical method is commonly applied to marketing strategy analysis [2–4]. The SWOT analytical method is able to help the decision maker of the enterprises evaluate qualitatively their competitiveness and can be used as a foundation of the development of strategies [5]. The quantitative SWOT such as external factor evaluation matrix (EFE), internal factor evaluation matrix (IFE), and competitive profile matrix (CPM) aim at analyzing statistical data, differing from the traditional SWOT analytical method [3, 6]. The disadvantage of the above approaches is that they cannot evaluate the qualitative and quantitative criteria simultaneously. Therefore, this paper proposes a new hybrid approach which integrates AHP method to carry out a complete evaluation of qualitative and quantitative criteria simultaneously and to evaluate the competitive environmental relationships between the several international distribution center locations within the ports in the asian region.

Erol and Ferrell Jr. [7] use fuzzy Quality Function Deployment (QFD) to convert qualitative information into quantitative parameters and then combine this data with other quantitative data to multiobjective mathematical programming model. Mikhailov and Tsvetinov [8] propose a fuzzy AHP approach for tackling the uncertainty and imprecision of the service evaluation process. Ronza et al. [9] present a quantitative risk analysis approach to port hydrocarbon logistics. Because risk is an uncertain criterion, it is not easy for a decision maker to measure exactly the value of risk. Chang and Huang [4] present a quantitative strength/weakness/opportunity/threat (SWOT) analysis method for assessing the competing strengths of major ports in East Asia. Yong [10] uses a fuzzy TOPSIS model to solve the problem of plant location choice. Önüt and Soner [11] propose an AHP/TOPSIS approach for solving transshipment site selection problem under fuzzy environment. Tahera et al. [12] develop a fuzzy logic approach for dealing with qualitative quality characteristics of a process. Chen et al. [13] combine fuzzy AHP with Multidimension Scaling (MDS) in identifying the preference similarity of alternatives. Chou et al. [14] present a new fuzzy multiple attributes decision-making (MADM) approach for solving facility location selection problem by using objective and subjective attributes. Lee and Lin [15] use a fuzzy quantitative SWOT procedure for environment evaluation of international distribution centers in Pacific Asian region. Lee et al. [16] introduce a fuzzy AHP and Balanced Score Card (BSC) approach for evaluating performance of IT department in the manufacturing industry in Taiwan. Chou [17] uses fuzzy MCDM approach for dealing with quantitative and qualitative criteria in a process of location choice. Chou [18] deals with objective data and subjective ratings by fuzzy logic. Chou [19] analyzes the competitive relationship between the ports of Hong Kong, Shanghai, and Kaohsiung by the sensitivity analysis.

In the past, although many researchers proposed a lot of fuzzy approaches, for example, fuzzy QFD, MADM, AHP, TOPSIS, BSC, SWOT, and MCDM, few presented a hybrid qualitative/quantitative fuzzy AHP model for dealing with both objective data and subjective criteria simultaneously in the decision-making process.

2.2. Fuzzy AHP

Despite of its wide application to various decision-making problems, the conventional AHP approach may not fully reflect a style of human thinking. Thus, the fuzzy AHP approach is proposed to overcome the disadvantage of the conventional AHP. The fuzzy AHP approach is a systematic method for the alternative choice and justification problems that combines the concept of fuzzy sets theory [20] and the hierarchical structure analysis [21].

Fuzzy AHP approach has been widely applied to many decision-making problems. For example, Chang [22] developed a fuzzy extent analysis for AHP and the approach is relatively easier in computational procedure than the other fuzzy AHP approaches. Kuo et al. [23] presented a fuzzy AHP method for the location choice of a convenience store. Kurttila et al. [5] combined AHP with SWOT to provide a new hybrid method for a forest certification case. Stewart et al. [24] combined AHP method with SWOT to present a new approach for improving the usability of AHP in strategic management. Kahraman et al. [25] applied fuzzy AHP to select the location of facility. Zhang et al. [26] combined fuzzy AHP with MCDM to deal with an MCDM decision-making problem. The results show that the proposed hybrid method was a useful way to deal with MCDM decision-making problems. Erensal et al. [27] determined key capabilities in technology management by using fuzzy AHP. Chan and Kumar [28] proposed a model for global supplier development considering risk factors by using fuzzy AHP. Bozbura and Beskese [29] determined the priorities of organizational capital measurement indicators by using fuzzy AHP. Bozbura et al. [30] used fuzzy AHP method to determine the priorities of human capital measurement indicators. Lee and Lin [15] developed a fuzzy quantified SWOT procedure that integrates MCDM concept and fuzzy AHP method for the location choice of international distribution centers.

3. Methodology

3.1. Fuzzy Sets Theory

Fuzzy sets theory is initially introduced by Zadeh [20]. A fuzzy number is defined as follows. Suppose 𝐴₁=(𝑐₁,𝑎₁,𝑏₁,𝑑₁) is a trapezoidal fuzzy number in Figure 1. The membership function of 𝐴₁ and the fuzzy arithmetic operations on fuzzy numbers are shown as follows: 𝑓_𝐴1(𝑥)=𝑥−𝑐₁𝑎₁−𝑐₁,𝑐₁≤𝑥≤𝑎₁1,𝑎₁≤𝑥≤𝑏₁𝑥−𝑑₁𝑏₁−𝑑₁,𝑏₁≤𝑥≤𝑑₁0,otherwise. (1)

Figure 1 

Trapezoidal fuzzy number 𝐴1=(𝑐1,𝑎1,𝑏1,𝑑1).

Suppose 𝐴₁=(𝑐₁,𝑎₁,𝑏₁,𝑑₁) and 𝐴₂=(𝑐₂,𝑎₂,𝑏₂,𝑑₂) are two trapezoidal fuzzy numbers.(a)Addition operation on𝐴₁and𝐴₂𝐴₁⊕𝐴₂=𝑐₁+𝑐₂,𝑎₁+𝑎₂,𝑏₁+𝑏₂,𝑑₁+𝑑₂. (2)(b)Subtraction operation on 𝐴₁ and 𝐴₂𝐴₁Θ𝐴₂=𝑐₁−𝑑₂,𝑎₁−𝑏₂,𝑏₁−𝑎₂,𝑑₁−𝑐₂. (3)(c)Multiplication operation on 𝐴₁ and 𝑟𝑟⊗𝐴₁=𝑟𝑐₁,𝑟𝑎₁,𝑟𝑏₁,𝑟𝑑₁. (4)(d)Division operation on 𝐴₁ and 𝑟𝐴₁𝑟=𝑐₁𝑟,𝑎₁𝑟,𝑏₁𝑟,𝑑₁𝑟. (5)

Chou [31] proposed the canonical representation of a triangular fuzzy number 𝑌=(𝑐,𝑎,𝑏): 𝑃(𝑌)=16(𝑐+4𝑎+𝑏). (6)

3.2. AHP Theory

Saaty [21] initially proposed the Analytic Hierarchy Process (AHP), which is a multiple-attribute decision-making tool for solving complex multiple-criteria decision-making problems. AHP methodology has some advantages. One of the most important advantages of the AHP is based on the pairwise comparison. Another is that the AHP calculates the inconsistency index, which is the ratio of the decision maker’s inconsistency on the criteria. The computational procedures for AHP are listed as follows.

Let us consider the criteria 𝐶₁,…,𝐶_𝑖,…,𝐶_𝑗,…,𝐶_𝑛,   someone level in the hierarchy. One wishes to find their weights of importance, 𝑊₁,…,𝑊_𝑖,…,𝑊_𝑗,…,𝑊_𝑛, on some elements in the next level. Allow 𝑎_𝑖𝑗,𝑖,𝑗=1,2,…,𝑛, to be the importance strength of 𝐶_𝑖 when compared with 𝐶_𝑗. In generally we can represent the comparative importance scale of criteria as shown in Table 2. The matrix of these numbers 𝑎_𝑖𝑗 is denoted by 𝐵: 𝑎𝐵=¹¹𝑎₁₂⋯𝑎_1𝑗⋯𝑎_1𝑛𝑎⋮⋮⋮⋮_𝑖1𝑎_𝑖2⋯𝑎_𝑖𝑗⋯𝑎_𝑖𝑛𝑎⋮⋮⋮⋮_𝑛1𝑎_𝑛2⋯𝑎_𝑛𝑗⋯𝑎_{𝑛𝑛𝑛×𝑛}, (7) where 𝑎_𝑗𝑖=1/𝑎_𝑖𝑗; that is, 𝐵 is reciprocal. If one’s judgment is perfect in all comparisons, then 𝑎_𝑖𝑘=𝑎_𝑖𝑗⋅𝑎_𝑗𝑘 for all 𝑖,𝑗,𝑘 and one calls the matrix 𝐵 consistent. An obvious case of a consistent matrix 𝐵 is its elements 𝑎_𝑖𝑗=𝑤_𝑖𝑤_𝑗,𝑖,𝑗=1,2,…,𝑛. (8)

Table 2 

Comparative importance scale of criteria.

Thus, when the matrix 𝐵 is multiplied by the vector formed by each weighting 𝑤=(𝑤₁,𝑤₂,…,𝑤_𝑛)^𝑇, one gets 𝑤𝐵𝑤=¹𝑤₁𝑤₁𝑤₂⋯𝑤₁𝑤_𝑗⋯𝑤₁𝑤_𝑛𝑤₂𝑤₁𝑤₂𝑤₂⋯𝑤₂𝑤_𝑗⋯𝑤₂𝑤_𝑛𝑤⋮⋮⋮⋮_𝑖𝑤₁𝑤_𝑖𝑤₂𝑤_𝑖𝑤_𝑗𝑤_𝑖𝑤_𝑛𝑤⋮⋮⋮⋮_𝑛𝑤₁𝑤_𝑛𝑤₂⋯𝑤_𝑛𝑤_𝑗⋯𝑤_𝑛𝑤_{𝑛𝑛×𝑛}𝑤₁𝑤₂⋮𝑤_𝑗⋮𝑤_𝑛𝑛×1𝑤=𝑛¹𝑤₂⋮𝑤_𝑗⋮𝑤_𝑛=𝑛𝑤. (9)

Because 𝑎_𝑖𝑗 is the subjective ratings given by the decision maker, there must be a distance between it and the actual values 𝑤_𝑖/𝑤_𝑗. Thus, 𝐵𝑤=𝑛𝑤 cannot be calculated directly. Therefore Saaty suggested using the maximum eigenvalue, 𝜆_max=(1/𝑛)(𝑤^′₁/𝑤₁+𝑤^′₂/𝑤₂+⋯+𝑤^′_𝑛/𝑤_𝑛), of the solution of matrix 𝐵 to replace 𝑛; then 𝐵𝑤=𝜆_max𝑤. (10)

By this method, one can obtain the characteristic vector, referred to as the priority vector. Besides Saaty suggested the consistency index (C.I=(𝜆_max−𝑛)/(𝑛−1)) and the consistency rate (C.R=C.I/R.I) to test the consistency of the intuitive judgment. In general, it is satisfactory and accepted if the value of C.I is about 0.1 and the value of C.R is less than 0.1.

3.3. Proposed Hybrid Fuzzy AHP Approach

The proposed hybrid fuzzy AHP approach to solving both quantitative data and qualitative ratings simultaneously in process of the location selection is introduced in this section. The proposed hybrid fuzzy AHP approach involves 11 steps shown as follows.

Step 1. Construct a hierarchical analysis structure in Table 3. These criteria in the hierarchical analysis structure can be divided into two categories: objective and subjective criteria. The objective criteria are defined in monetary or quantitative terms. The subjective criteria are defined in linguistic terms represented by fuzzy numbers.

Table 3 

A hierarchical analysis structure.

Step 2. Introduce linguistic variables for importance weight of criteria. Terms of linguistic variables for importance weight of criteria could be called “equally important”, “weakly important,” “strongly important,” “demonstrably important,” “absolutely important,” and so forth. These linguistic variables can be expressed in fuzzy numbers such as “equally important” = (1,1,1), “weakly important” = (2,3,4), “strongly important” = (4,5,6), “demonstrably important” = (6,7,8), and “absolutely important” = (9,9,9). Their reciprocals are considered as “weakly unimportant” = (1/4,1/3,1/2), “strongly unimportant” = (1/6,1/5,1/4), “demonstrably unimportant” = (1/8,1/7,1/6), and “absolutely unimportant” = (1/9,1/9,1/9).

Step 3. Introduce linguistic variables for ratings of alternative locations. Terms of linguistic variables for ratings of alternative locations could be called “very poor,” “poor,” “fair,” “good,” “very good,” and so forth. These linguistic variables can be expressed in fuzzy numbers such as “very poor” = (0.0,1.0,2.0), “poor” = (1.0,2.0,3.0), “fair = (2.0,3.0,4.0)”, “good” = (3.0,4.0,5.0), “very good” = (5.0,5.0,5.0), and so forth.

Step 4. Determine the importance weights of criteria by the decision maker. Assume there are 𝑁 candidate locations (𝐴₁,𝐴₂,𝐴_𝑛,…,𝐴_𝑁),𝐼 evaluation criteria (𝐶₁,𝐶₂,𝐶_𝑖,…,𝐶_𝐼), and 𝐽 subcriteria (𝐶_𝑖1,𝐶_𝑖2,𝐶_𝑖𝑗,…,𝐶_𝑖𝐽) under criteria 𝑖, where 1≤𝑛≤𝑁,1≤𝑖≤𝐼,1≤𝑗≤𝐽. 𝑊_𝑖=(𝑐_𝑖,𝑎_𝑖,𝑏_𝑖) and 𝑊_𝑖𝑗=(𝑐_𝑖𝑗,𝑎_𝑖𝑗,𝑏_𝑖𝑗) are the fuzzy importance weights given by the decision maker to criteria 𝐶_𝑖 and subcriteria 𝐶_𝑖𝑗, respectively.

Step 5. Defuzzify the weights of criteria and subcriteria. Then calculate the normalized weights. According to (6) proposed by Chou [31], we can obtain the representation of fuzzy numbers 𝑊_𝑖 and 𝑊_𝑖𝑗 as follows: 𝑃𝑊_𝑖=16𝑐_𝑖+4𝑎_𝑖+𝑏_𝑖,𝑃𝑊_𝑖𝑗=16𝑐_𝑖𝑗+4𝑎_𝑖𝑗+𝑏_𝑖𝑗. (11) The normalized weights of criteria 𝐶_𝑖 and subcriteria 𝐶_𝑖𝑗 are given by 𝑊_𝑖=𝑃𝑊_𝑖∑^𝐼_𝑖=1𝑃𝑊_𝑖,𝑊_𝑖𝑗=𝑃𝑊_𝑖𝑗∑^𝐽_𝑗=1𝑃𝑊_𝑖𝑗, (12) where ∑^𝐼_𝑖=1𝑊_𝑖=1, ∑^𝐽_𝑗=1𝑊_𝑖𝑗=1.

The weight vector is therefore formed as follows: 𝑊=𝑊₁×𝑊₁₁,𝑊₁×𝑊₁₂,…,𝑊₁×𝑊_1𝑗,…,𝑊_𝑖×𝑊_𝑖1,𝑊_𝑖×𝑊_𝑖2,…,𝑊_𝑖×𝑊_𝑖𝑗,…,𝑊_𝐼×𝑊_𝐼1,𝑊_𝐼×𝑊_𝐼2,…,𝑊_𝐼×𝑊_𝐼𝐽=𝑊₁₁₁,𝑊₁₁₂,…,𝑊_11𝑗,…,𝑊_𝑖𝑖1,𝑊_𝑖𝑖2,…,𝑊_𝑖𝑖𝑗𝑊,…,_𝐼𝐼1,𝑊_𝐼𝐼2,…,𝑊_𝐼𝐼𝐽. (13)

Step 6. Calculate the maximum eigenvalue (𝜆_max), the consistency index (C.I), and the consistency rate (C.R) for AHP model to test the consistency of the intuitive judgment.

Step 7. The decision maker assesses alternatives under subjective criteria. Let 𝑅_𝑖𝑗𝑛=(𝑐_𝑖𝑗𝑛,𝑎_𝑖𝑗𝑛,𝑏_𝑖𝑗𝑛) be the fuzzy ratings given by the decision maker to alternative 𝐴_𝑛 under subjective subcriteria 𝐶_𝑖𝑗.

Step 8. Assess alternatives under objective criteria. Let 𝑅^𝑜_𝑖𝑗𝑛=(𝑐^𝑜_𝑖𝑗𝑛,𝑎^𝑜_𝑖𝑗𝑛,𝑏^𝑜_𝑖𝑗𝑛) be the fuzzy quantity given to alternative𝐴_𝑛 under objective sub-criteria𝐶_𝑖𝑗. The objective criteria are determined in various units and must be transformed into dimensionless indices (or ratings) to ensure compatibility with the linguistic ratings of subjective criteria. The alternative with the minimum cost (or maximum benefit) should have the highest rating. By (14) and (15), we can transform fuzzy quantities for objective subcriteria into fuzzy ratings: 𝑅_𝑖𝑗𝑛=𝑅^𝑜_𝑖𝑗𝑛max_𝑖𝑏^𝑜_𝑖𝑗𝑛×5, (14) where max_𝑖{𝑏^𝑜_𝑖𝑗𝑛}>0 and 𝑅_𝑖𝑗𝑛 denotes the transformed fuzzy rating of objective fuzzy benefit 𝑅^𝑜_𝑖𝑗𝑛. 𝑅_𝑖𝑗𝑛 becomes larger when objective fuzzy benefit 𝑅^𝑜_𝑖𝑗𝑛 is larger: 𝑅_𝑖𝑗𝑛=min_𝑖𝑐^𝑜_𝑖𝑗𝑛𝑅^𝑜_𝑖𝑗𝑛×5, (15) where min_𝑖{𝑐^𝑜_𝑖𝑗𝑛}>0 and 𝑅_𝑖𝑗𝑛 denotes the transformed fuzzy rating of objective fuzzy cost 𝑅^𝑜_𝑖𝑗𝑛. 𝑅_𝑖𝑗𝑛 becomes smaller when objective fuzzy cost 𝑅^𝑜_𝑖𝑗𝑛 is larger.

Step 9. Construct a fuzzy rating matrix based on fuzzy ratings. The fuzzy rating matrix𝑀can be concisely expressed in matrix format: 𝑅𝑀=¹¹¹𝑅₁₂₁⋯𝑅_𝐼𝐽1𝑅₁₁₂𝑅₁₂₂⋯𝑅_𝐼𝐽2𝑅∶∶⋯∶_11𝑁𝑅_12𝑁⋯𝑅_𝐼𝐽𝑁. (16)

Step 10. Obtain the total fuzzy rating (𝑅) based on the fuzzy rating matrix (𝑀) and weight vector (𝑊): 𝑅𝑅=¹¹¹𝑅₁₂₁⋯𝑅_𝐼𝐽1𝑅₁₁₂𝑅₁₂₂⋯𝑅_𝐼𝐽2𝑅⋮⋮⋯⋮_11𝑁𝑅_12𝑁⋯𝑅_𝐼𝐽𝑁⊗𝑊₁₁₁𝑊₁₁₂⋮𝑊_𝐼𝐼𝐽=𝑅₁𝑅₂⋮𝑅_𝑁. (17)

Step 11. Defuzzify the total fuzzy rating by (6) and then rank alternatives according to their total crisp ratings. Finally, we can select easily the best alternative with the maximum total crisp ratings: 𝑃𝑅_𝑛=16𝑐_𝑛+4𝑎_𝑛+𝑏_𝑛. (18)

4. A Case Study

In this section, the proposed hybrid fuzzy AHP approach is applied to the location choice of international distribution centers in the global logistics of the multinational corporation. A Taiwanese multinational corporation plans to select an appropriate location of international distribution center at the international transshipment port. After initial screening, three alternative port locations including the port 𝐴₁, the port 𝐴₂, and the port 𝐴₃ are selected for further evaluation. The procedures for evaluation are shown as follows.

Step 1. Constructing a hierarchical analysis structure is shown in Table 3. There are 6 criteria and 30 subcriteria in the hierarchical analysis structure summarized by Chou [19]. These criteria in the hierarchical analysis structure can be divided into two categories: objective and subjective criteria. The objective criteria include proximity of the feeder port (𝐶₁₂), frequency of ship calls (𝐶₁₄), port charge (𝐶₂₅), volumes of import containers (𝐶₃₁), volumes of export containers (𝐶₃₂), volumes of transshipment containers (𝐶₃₃), port facilities (𝐶₆₁), loading and unloading facilities (𝐶₆₂), and cargo handling efficiency (𝐶₆₄). The others are subjective criteria. The objective criteria are defined in quantitative terms (e.g., nautical mile, $US, TEU). The subjective criteria are defined in linguistic terms represented by fuzzy numbers.

Step 2. Present the linguistic variables and fuzzy numbers for comparative importance weights of criteria.

Step 3. Present the linguistic variables and fuzzy numbers for ratings of alternatives.

Step 4. Determine the fuzzy comparative importance weights of criteria and subcriteria by the decision maker in Tables 4 and 5, respectively.

Table 4 

Survey data for fuzzy comparative weights for each criterion.

Table 5 

Survey data for fuzzy comparative weights for each subcriterion.

Step 5. Defuzzify the fuzzy weights of criteria and subcriteria. Then calculate the normalized weights in Table 3.

Step 6. Calculate the maximum eigenvalue (𝜆_max), the consistency index (C.I), and the consistency rate (C.R) for AHP model to test the consistency of the intuitive judgment:11𝐴𝑤=413322141454431411213131512111121411113214=,𝜆31310.15270.43940.08240.07570.07850.17140.97932.75740.53010.50550.47991.1449_max=160.9793+0.15272.7574+0.43940.5301+0.08240.5055+0.07570.4799+0.07851.1449𝜆0.1714=6.4331,C.I=^max−𝑛=𝑛−16.4331−66−1=0.08,C.R=C.I=R.I0.081.24=0.07. (19)

Step 7. The decision maker assesses alternatives under subjective criteria. For example, the fuzzy ratings given by the decision maker to the port 𝐴₁ under subjective subcriteria are shown in Table 3.

Step 8. Assess alternatives under objective criteria. The fuzzy quantities given to alternative under objective subcriteria are listed in Table 6. The objective fuzzy quantities are determined in various units (e.g., nautical mile, $US, TEU) and must be transformed into dimensionless indices (or ratings) to ensure compatibility with the linguistic ratings of subjective criteria. By (14) and (15), we can transform fuzzy quantities for objective subcriteria into fuzzy ratings in Table 6.

Table 6 

Fuzzy ratings for alternatives under objective criteria.

Step 9. Construct a fuzzy rating matrix based on fuzzy ratings.

Step 10. Obtain the total fuzzy ratings based on the fuzzy rating matrix and the weight vector.

Step 11. By (6), we can defuzzify the total fuzzy ratings and the total crisp ratings for the port 𝐴₁, the port 𝐴₂, and the port 𝐴₃ are 3.72, 3.88, and 4.41, respectively. Finally, the decision maker of Taiwanese multinational corporation selects easily the port 𝐴₃ with the maximum total crisp ratings as the best location for international distribution center in the global logistics system.

5. Conclusions

The paper proposes a new hybrid fuzzy AHP model for dealing with both objective and subjective criteria in process of decision making simultaneously. The proposed hybrid fuzzy AHP model is applied to solve the location choice problem of international distribution center in the global logistics of multinational corporation. The results show that the proposed new hybrid fuzzy AHP model is an appropriate and more efficient approach to deal with both objective and subjective criteria in process of decision making simultaneously. The hybrid fuzzy AHP model in this paper overcomes the disadvantages of quantitative or qualitative approaches in the previous literature. The proposed hybrid fuzzy AHP approach can not only solve the problems of location choice, but also many other decision-making problems.

Acknowledgment

This research work was partially supported by the National Science Council of Taiwan under Grant no. NSC 101-2514-S-022-001.