Suppliers Selection Using FAHP and FTOPSIS in a Chemical Manufacturing Company

The research aimed to determine the appropriate weight criteria and sub-criteria in selecting the chemical solvent supplier. It was also to recommend the best alternative suppliers using a comparative analysis method of Fuzzy Analytical Hierarchy Process (FAHP) and Fuzzy Technique for Order of Preference by Similarity to Ideal Solution (FTOPSIS). The research was conducted in one of the chemical manufacturing companies in Indonesia. The type of research was descriptive with a quantitative approach that used primary data sources such as questionnaires, interviews, and observations, and secondary data such as books and journals. The criteria studied were cost, quality, delivery, flexibility, supplier profile, and supplier relationship. The results of using FAHP indicate that the criteria with a major influence on the decision making of supplier selection are quality followed by delivery, cost, supplier relationship, flexibility, and supplier profile in sequence. Then, using FTOPSIS, it recommends that the best supplier alternative decision is Supplier A with a weight value of 0,773 and an approved evaluation status. The results have shown that both methods are suitable for supplier selection, particularly in supporting group decision making and modeling uncertainty.


INTRODUCTION
Indonesia's economic structure by business in 2017 was still dominated by business fields, namely the manufacturing or industrial sector (Kementerian Perindustrian Republik Indonesia, 2019). The researchers study the company engaged in that industry, which is a chemical manufacturing company. According to interviews with the Operational Manager, the number of products owned by the company certainly requires the company to have several suppliers to meet the needs of raw materials in which each of it has several suppliers. The variety of suppliers and its criteria makes the company face difficulties in choosing the best suppliers to cut expenses and to achieve profit targets. It is necessary to select the right supplier to minimize the comparison of market prices and the difference in the number of raw materials and obtain the quality of raw materials and other aspects that will optimize Supply Chain Management (SCM) in this industry.
The term of SCM was credited to Oliver and Webber in 1982 and was quickly introduced to academia by Jones andRiley in 1987 andEllram andCooper in 1990 (Carter, Rogers, & Choi, 2015). It consists of all parties involved in meeting customer demands directly or indirectly. It does not only involve producers and suppliers, but also transporters, warehouses, retailers, and customers. In every organization, such as a manufacturer, it provides all involved functions in receiving and filling customers' requests. These functions are not limited I N P R E S S to new product development, marketing, operations, distribution, finance, and customer service (Chopra & Meindl, 2016). According to Jain, Sangaiah, Sakhuja, Thoduka, and Aggarwal (2018), supplier selection is a strategic decision that companies take as a result of which they identify and evaluate potential suppliers offering high-value products and services. The rapid advances and competition between companies need effective management of the supply chain for gaining more profit. Similarly, Junior, Osiro, and Carpinetti (2014) agreed that supplier selection was one of the most important activities of acquisition as its results had a great impact on the quality of goods and performance of organizations and supply chains. Through supplier selection, it was also possible to anticipate the evaluation of the potential of suppliers to establish a collaborative relationship.
In selecting suppliers, it is necessary to determine the criteria as a material for the selection of alternative suppliers. Supplier selection criteria can form a set of strategic qualifications and operational standards used by buyers to align between external sources and internal goals. The right supplier selection criteria can help guarantee the uncertainty in the supplier's role and the possibility to predict outcomes and responses (Nair, Jayaram, & Das, 2015). The criteria can vary depending on the type of product or industry being considered (Gurung & Phipon, 2016). To make accurate supplier selection and appropriate criteria, the company requires methods that can support decision making. One of the methods for decision making is Multi Criteria Decision Making (MCDM). According to Mulliner, Malys, and Maliene (2016), MCDM is a set of methods for evaluating alternatives in terms of numerous conflicting decision criteria. Thus, given a set of alternatives (options) and several decision criteria, the goal of MCDM is to provide a choice, ranking, description, classification, sorting, and order of alternatives from the most preferred to the least preferred option. According to Taylan, Bafail, Abdulaal, and Kalbi (2014), in MCMD problems, the attribute values and the relative weights are usually characterized by a fuzzy number.
If the value assigned is 0, the element does not belong to the set (it has no membership). If the value assigned is 1, the element belongs entirely to the set (it has no membership). Finally, if the value lies within the interval [0, 1], the element has a certain degree of membership (it belongs partially to the fuzzy set). Then, a fuzzy set contains elements that have different degrees of membership in it. The main characteristic of fuzzy is the grouping of individuals into classes that do not have sharply defined boundaries. Based on Junior et al. (2014), fuzzy set theory combined with MCDM methods has been extensively used to deal with uncertainty in the supplier selection decision process, since it provides a suitable language to handle imprecise criteria and to integrate the analysis of qualitative and quantitative factors. From many methods of MCDM, the researchers choose the most popular methods of supplier selection. It is fuzzy set theory combined with MCDM methods such as Fuzzy Analytical Hierarchy Process (AHP) and Fuzzy  Technique for Order of Preference by Similarity to  Ideal Solution (TOPSIS). FAHP is the perfecting method of the AHP method. It covers the weaknesses that exist in the AHP method. According to Thomas L. Satty in Devi and Wardhana (2018), AHP is a method used in the decision-making process of complex problems. Problems solved using the AHP method are said to be complex if the structure of the problem is inaccurate. So, the input used to solve this problem is human thought. The AHP method breaks a situation into parts and organizes these parts or variables into a hierarchical arrangement.
Meanwhile, TOPSIS was first introduced by Yoon and Hwang in 1981. They developed the TOPSIS method based on the concept that the alternative chosen had to have the shortest distance from the positive-ideal solution and the longest distance from the negative ideal solution (Lu, Zhang, Ruan, & Wu, 2007). According to Ertugrul and Karakasoglu in Chatterjee and Stević (2019), by using FAHP and FTOPSIS, uncertainty and vagueness from subjective perception and the experiences of decision-maker can be adequately represented, and it can reacha more effective decision. FAHP is employed to determine the relative importance of the selected criteria. Then, based on the points given to the alternatives in different selected criteria by the personnel interviewed, FTOPSIS is used to rank the competing suppliers (Sultana, Ahmed & Azeem, 2015). Özbek (2015) suggested that the supplier selection model was developed to help managers of a company make decisions in selecting suppliers. The researcher used FTOPSIS. The results showed that the most appropriate supplier was supplier B. However, there was no significant difference between suppliers, especially Suppliers A and C. The company could choose to work with these three suppliers, but Supplier B was the top priority.
Moreover, Junior et al. (2014) analyzed comparative studies of FAHP and FTOPSIS methods applied to supplier selection. Despite the large number of articles proposing the use of FAHP and FTOPSIS, they said there were no comparative studies of these two methods applied to supplier selection. The results revealed that both methods were suitable for supplier selection, particularly to support group decision making and model uncertainty. However, the comparative analysis in this research had shown that the FTOPSIS method suited the supplier selection regarding changes of alternatives and criteria, agility and number of criteria, and alternative suppliers. Li et al. (2019) studied the distribution network of supplier selection with FAHP and TOPSIS. They said choosing the proper suppliers for distribution network commodities was very important. Due to the stringent requirement on the safe and normal operation, the suppliers for distribution network commodities were assessed by various criteria. They I N P R E S S found out that cost, quality, delivery, and service were the most important criteria. Similarly, Karabayir, Botsali, Kose, and Cevikcan (2019) analyzed the supplier selection method in a construction company in Turkey using the FAHP and FTOPSIS. In the past, the price list was sufficient for choosing the supplier, but nowadays, selection depended on numerous criteria such as price, quality, delivery, variety, and warranty. Later on, they indicated six commonly used criteria for supplier evaluations. The criteria obtained were flexibility, quality, price, relationship, profile, and delivery. The result showed that the value of criteria was different based on both methods, but the rankings were similar. The ranking of criteria was the price, delivery, relationship, profile, flexibility, and quality, respectively, according to the results of FTOPSIS.
Meanwhile, according to FAHP, the criteria weights were sorted to price, delivery, profile, relationship, flexibility, and quality. The diversity in relationship and profile criteria weights was mainly due to the differences in algorithms. For the sequence of alternatives results, the first preferred supplier was A2, according to both methods.
Different from the others, Nazari-Shirkouhi, Miri-Nargesi, and Ansarinejad (2017) examined project selection with FAHP and FTOPSIS. They proposed a structured method for outsourcing information systems for project selection using FAHP and FTOPSIS analysis with seven criteria and five alternative decisions. To accommodate the criteria, FAHP was chosen to get the relative weights of the criteria. It was found that FTOPSIS was more practical for ranking information systems in terms of their overall performance concerning several criteria. In addition to these two methods, several methods that had been extended in the fuzzy environment could be used to compare the results used. Chen, Chou, Luu, and Yu (2016) said that in sorting the importance of the FAHP criteria, it had a lower complexity time compared to other fuzzy sets methods. They showed that the calculations were easier. In accordance with the previous statement, Junior et al. (2014) also revealed that FAHP was an efficient tool for dealing with data inconsistencies involved in deciding the preferences of different decision variables. As for the FTOPSIS, Simić, Kovačević, Svirčević, and Simić (2017) stated that FTOPSIS had no limitation of increasing the number of alternative suppliers. So, any increase in the number of alternative suppliers would not affect the calculation results, and FTOPSIS delivered more consistent results. However, alternative input was not optimal.
Based on some previous research, the research focuses on measuring the weight of criteria and providing a priority level for alternative suppliers using FAHP. The research also evaluates a priority level based on the distance to the ideal solution, both positive and negative using FTOPSIS. The research focuses on a chemical manufacturing company in Indonesia. It is located at Cakung, East Jakarta City.
The most top-selling product is called as OD 661. Figure 1 displays the three most purchased materials, which are raw materials for OD 661 products. The criteria used are the combination of the adaptation from several previous researchers that suit the needs of the company. Those criteria are summarized in Table 1.

METHODS
A descriptive quantitative approach is chosen as the research method, and the time horizon is crosssectional with a chemical manufacturing company as the unit of analysis. The research uses both primary and secondary data. The primary data are from questionnaires, interviews, and observations. Interviews are conducted by the researchers with two respondents (Operational Manager and Purchasing I N P R E S S Manager) as an expert in the company. Meanwhile, the secondary data used are the pages of government publications, previous literature review, data of purchasing raw materials, books, and journals. The research uses a comparative method between FAHP and FTOPSIS. In the analysis process, both methods have some similarities on the steps. First, it is specifying the supplier selection criteria and subcriteria indicators. Second, the researchers interview the responsible source in the purchasing department in the company to collect alternative suppliers. Third, it is making the questionnaire that contains the assessment weights for criteria and sub-criteria and gives it to the responsible source. Fourth, the researchers calculate the results from the questionnaire for the weight of criteria, sub-criteria, and alternatives. Last, from the result, researchers receive the final level of priorities, conclusions, and suggestions.

RESULTS AND DISCUSSIONS
The researchers examine the best-selling product of the company that is OD 661 consisting of three chemical compounds. Each compound has two alternative suppliers. The compounds are surfactant, glycol, and detergent. The researchers decide to calculate these compounds because they are the most frequently purchased by the company. In this research, the surfactant will be the calculation example. The calculation is divided into FAHP and FTOPSIS. In FAHP, the first step is to create a hierarchical model based on criteria, sub-criteria, and alternatives as  Figure 2. The second step of FAHP is to enter the weight of the questionnaire into the fuzzy scale. Then, the researchers convert it into the Triangular Fuzzy Number (TFN), which is followed by pairwise comparisons. Tables 2 to 4 show the fuzzy scale, TFN used, and the pairwise comparison between criteria sequently.  The third step of FAHP analysis is to calculate the factors of paired comparison matrices. Those have been converted into TFN using the geometric mean calculation method adapted from Huynh et al. (2018). The equations are as follows: Based on the result in Table 4 (see appendices), the researchers convert the value to Table 5. Is the reverse value which is calculated using Equation (3). The example is shown in the following calculation using Equation (1). 8,9,9) × (1,1,1) × (1,1,1)] 1/3 = (1.074, 1.080, 1.082) The fourth step is converting the value in Table  5 to the value in Table 6. The value in Table 6 is the relative fuzzy weight. The equations and the example of the relative fuzzy weight of each criterion (w i ) are as follows: The fifth step is to calculate normalized weights as shown in Table 7. The normalization weight determines the pairwise comparison matrix ranking of each comparison. The highest normalization weights are taken for the final decision. The normalized weightsare calculated by using the following formula:   Based on the steps mentioned, the final result of the FAHP in the supplier selection process of the surfactant is in Tables 8−10. From the results of the comparison of importance between the criteria using the FAHP in Table 8, it is found that the quality has the highest importance level with a weight of 0,231. Then, it is followed by delivery (0,197), cost (0,179), supplier relationship (0,136), flexibility (0,132), and supplier profile (0,125).  The results of the importance comparison between sub-criteria using the FAHP in Table 9 obtain the three best sub-criteria of each criterion. In the cost criteria, the first priority is the price. Then, the second and third priorities are the quantity and logistic cost respectively. For the quality criteria, the order is after-sale service, rejection percentage, and product reliability. Then, in the delivery, there are safety and security, the quantity of conformance, and on-time delivery. Next, for the flexibility criteria, the order is quantity adjustment, product specifications, and time adjustment. In the supplier profile criteria, there are supplier reputation and financial health. Last, the supplier relationship criteria have trust, cooperation, and reputation for integrity.
Based on Table 10, the results of the FAHP show that Supplier A is the best supplier with a value of 23,157 for surfactant. The difference of 1,299 is higher than Supplier B. To summarize the calculation results from the FAHP, it has found the prioritized criteria, sub-criteria, and the best alternative supplier. Thus, the researchers can recommend the supplier, criteria, and sub-criteria to be considered by the company in the process of supplier selection for surfactant.
Next, it is FTOPSIS application. The first step of using FTOPSIS is to input the weight of the questionnaires into the TFN based on the linguistic scale and the importance criteria from the questionnaires. The results are shown in Tables 11−13.  Table 11 shows the linguistic scale for evaluating the rank of interest criteria. VL is denoted by TFN value of 0,0; 0,0; and 0,25. For M, it is denoted by TFN value of 0,0; 0,25; and 0,50. For I, it is shown by TFN value of 0,25; 0,50; and 0,75. For VI, it is denoted by the TFN value of 0,50; 0,75; and 1,0. Last, in AI, there is the TFN value of 0,75; 1,0; and 1,0. Table 12 presents the linguistic scale to evaluate alternative supplier ratings. Similar to Table 11, it shows the TFN values of each linguistic scale. For example, VL is denoted by 0,0; 0,0; and 2,5.    The first step that must be done is to sum the results of the questionnaire data comparison between the criteria and sub-criteria with the alternative suppliers. Table 13 is a recapitalization of the questionnaires for the importance of the criteria. The same summary is done in the alternative supplier questionnaires.

I N P R E S S
The second step of using FTOPSIS is to change the questionnaire data that have been recapitulated into a fuzzy scale. The results can be seen in Table 14. The weight of the questionnaire is converted into TFN based on Table 11. The number is based on the weight of the questionnaire. In Table 14, it can be seen that the criteria for C1 in respondent 1 obtain the value of 0,75; 1,00; and 1,00. According to the recapitulation in Table 13, the C1 criteria in respondent 1 is AI. Then, for alternative supplier conversion, it can be seen from the recapitulation of the questionnaire with fuzzy scale weights based on Table 12. For example, if Supplier A has good questionnaire weight (G), based on Table 12, weights of G fuzzy scale are 2,5; 5,0; and 7,5.  Then, for alternative supplier conversion, it is seen from the recapitulation of questionnaires with fuzzy scale weights based on Table 15. It is the result of the calculation of criteria weight (Wij). The point of l means it is filled with the lowest weight value by respondents 1 and 2. Then, the point of m implies it is medium. Moreover, the point of u shows it has the biggest weight of respondents 1 and respondent 2. The third step of FTOPSIS is to calculate the weight of sub-criteria (W j ) by using the following formula:

I N P R E S S
The fourth step of FTOPSIS is to calculate the combination matrix of respondents 1 and 2 using Equation (8). In line with the explanation Table  15, Table 16 is the result of a matrix combination formulation. The point of l means the lowest weight value of respondent 1 and 2. Then, the point of m implies it is medium using Equation (9). Moreover, the point of u shows it has the biggest weight of respondents 1 and 2. To convert the value in Table 16, the next step of FTOPSIS is to calculate normalized fuzzy decision matrix (r ij ). The result of normalized fuzzy decision matrix is shown in Table 17. The normalized fuzzy decision is calculated by using the following formula. ij ] m × n (9) ij ij /u j + , ij /u j + , ij /u j Before calculating the r ij , researchers first identify each criterion with the benefit criteria and cost criteria categories. After the identification, the calculation is done. The u ci + of 10,0 is obtained from the largest number of fuzzy numbers C1. The example of the calculation is described as follows:  The sixth step of FTOPSIS is to calculate the weight of normalized fuzzy (Vij) by using Equations (13) and (14). Table 18 is the result of the calculation of weight normalized fuzzy calculation. The example is in Equation (15) V c1 (supplier A) = (0,75×0,50), (1,00×0,88), (1,00×1,00) = (0,38, 0,88, 1,00) The sixth step of FTOPSIS is to calculate Fuzzy Positive Ideal Solution (FPIS) (A+) and Fuzzy Negative Ideal Solution (FNIS) (A-) by using Equations (16) and (17). Table 19 is the result of the calculation of FPIS and FNIS. The A+C1 is obtained from the largest value (max) of V C1 at Supplier A and Supplier B. Then, the A-C1 is from the lowest value (min) of V C1 at Supplier A and Supplier B. The V C1 of Supplier A is 0,38; 0,88; and 1,00. Moreover, V C1 of Supplier B is 0,00; 0,38; and 1,00. So, A+C1 of l, m, and u is 0,38; 0,88; and 1,00, and A-C1 is 0,00; 0,38; and 1,00. The seventh step of FTOPSIS is to calculate the distance of d+ and d-from each alternative by using Equations (18) and (19). Table 20 shows the results of d + and d-. The d+ can also be interpreted as the distance of a positive ideal solution. Meanwhile, dis the distance of a negative ideal solution. This step aims to find the range of positive and negative ideal solutions from each sub-criterion.
I N P R E S S     Based on the steps done, the final result of the fuzzy TOPSIS calculation in the supplier selection for surfactant is done. In Table 23, it can be seen that Supplier A is the first priority with a weight of 0,773. Thus, the evaluation status is approved. Meanwhile, Supplier B with 0,292 is recommended but with a high risk. It can be said that if the company continues to work with Supplier B, there will be the possibility of inefficiencies and inadequacies that can harm the company. This is inline with glycol and detergent result, which Supplier A become the first priority in Table 23 (see appendices).

CONCLUSIONS
According to the results of calculations using the FAHP and FTOPSIS, Supplier A is the first priority with an approved evaluation status. Meanwhile, Supplier B is recommended but with a risk. This conclusion is obtained from the calculation results of the suppliers that have the highest priority order using FAHP method, the shortest distance from the positive ideal solution, and the longest distance from the negative ideal solution using calculations with the FTOPSIS. The calculations using the FTOPSIS have advantages in the form of weight results with a scale of preference recommendations, and they indicate the level of risk to be faced. This method also considers the distance of a positive ideal solution which can provide decisions in maximizing corporate profits. Moreover, FAHP also functions well to identify which priorities should be prioritized in weighting the criteria and subcriteria.
The research concludes that the combination of fuzzy sets theory with MCDM is a suitable approach in supplier selection. It is because the AHP method is very suitable in determining the weighting of criteria and selection of alternative suppliers with priority weighting and TOPSIS in determining supplier alternative selection with a positive ideal solution distance. Then, fuzzy sets get maximum results by eliminating ambiguity and inconsistencies from uncertain criteria and sub-criteria. This approach is very suitable for companies to produce unanimous decisions. Based on that consideration, the research can enrich the body of knowledge to provide additional deficiencies in previous research. It is expected that the future research can be further developed using a combination of rarely found methods.