Modern global markets are characterized by numerous factors, among which globalization, rising customer value expectations, expanding regulatory compliance requirements, the global economic crisis, and intense competitive pressure are of paramount importance [1]. Consequently, in response to these factors, manufacturing companies must first select and then retain their key suppliers. Selecting an appropriate supplier can significantly reduce an organization's purchasing costs, enhance market competitiveness, and increase end-user satisfaction. Therefore, the purchasing and procurement unit plays a critical role in an organization's efficiency and effectiveness. The performance of this unit directly impacts cost reduction, profitability, and organizational flexibility [2]. Undoubtedly, the most important and sensitive stage in any organization's purchasing process is supplier evaluation and selection [3].

The importance of supplier selection stems from the fact that suppliers commit to providing resources while simultaneously influencing activities such as inventory management, production planning and control, cash flow requirements, and product quality. As the importance of purchasing increases, purchasing decisions become more critical; moreover, as companies grow increasingly dependent on their suppliers, the direct and indirect consequences of poor decision-making in this area become more apparent [4].

In the supplier selection process, evaluation is conducted by assessing various criteria and supplier attributes. Numerous studies have been conducted on supplier selection, in most of which supplier selection is treated as a Multi-Criteria Decision-Making (MCDM) process [5]. Accordingly, various MCDM methods have been employed to address the supplier selection problem. The main limitation of MCDM approaches is their inability to manage risks throughout the decision-making process effectively. However, in risk environments, decision-makers always seek to manage risk; thus, in such contexts, conventional MCDM approaches are ineffective for risk management [6].

Risk can be appropriately managed using the Failure Mode and Effects Analysis (FMEA) method. FMEA is a systematic process for identifying potential failures and is applied during the design and implementation stages of a process before failures occur. The objective of this method is to eliminate or reduce potential failures [7]. Essentially, FMEA is a reliable evaluation technique used to identify failure modes that may reduce a system's overall reliability. Functional analyses are used as inputs to determine valid failure modes across all system levels. FMEA is applied to reduce risk generation (by decreasing the severity of failure effects), reduce the likelihood of failure, or achieve both objectives [8]. Fundamentally, FMEA is an inductive analysis technique.

Nevertheless, the probability of failure can only be determined or reduced through an understanding of failure occurrence mechanisms. Ideally, by eliminating the causes (roots) of failure, the probability of occurrence can be reduced to a level approaching "impossible." Therefore, in FMEA, extensive information regarding failure causes (deductive analysis) must be prepared [9].

However, the primary limitation of the FMEA technique lies in its inability to assign weights to decision criteria. In real-world problems, each criterion and sub-criterion carries a different weight. If the weights of decision criteria are not calculated accurately, the decision-maker may be led toward incorrect conclusions. Hence, precise measurement of the weights of criteria and sub-criteria in the decision-making process is unavoidable. In traditional FMEA, different combinations of severity, detectability, and occurrence can yield identical Risk Priority Numbers (RPNs), even though the actual risk consequences may differ. As a result, when multiple identical RPN values are generated from different criteria, decision-making becomes extremely difficult. Therefore, based on existing studies, the use of traditional FMEA and conventional MCDM methods for supplier selection in a risk environment is not sufficiently effective. In such environments, both the risk associated with decision criteria and their relative weights must be accurately calculated.

Buckley's Fuzzy Analytic Hierarchy Process (F-AHP) is a powerful and flexible method for solving MCDM problems under uncertainty [10]. This method offers several advantages, including flexibility, the ability to address uncertainty, and the capability to integrate knowledge from multiple stakeholders. Nevertheless, Buckley's F-AHP also has limitations, such as computational complexity and the need for specialized software. F-AHP can assist in selecting the best supplier by considering quality, price, and reliability in a fuzzy context.

In this study, the authors employ an integrated FMEA approach that addresses the existing limitations. The main objective of this research is to address research gaps in integrating FMEA and MCDM methods. Unexpected downtime in production lines, frequent failures of critical equipment, sudden declines in sales, and disruptions to work processes are undesirable events that can cause serious, and sometimes irreparable, damage to an organization. Senior managers are constantly seeking techniques that can deliver optimal results in the shortest possible time, enabling them to predict high-risk events before they cause severe losses and to take corrective actions accordingly. The purpose of applying the FMEA method is to take appropriate and effective actions to eliminate or reduce failures based on their priority. Moreover, analyzing failure modes can document the current status of activities and their failure probabilities, providing valuable records for future processes.

Fuzzy set theory was introduced by Zadeh [11] to eliminate and neutralize the imprecision inherent in human judgments. The F-AHP can systematically solve selection problems that utilize concepts from fuzzy set theory [12]. Applications of F-AHP have been reported in various fields, including the selection of energy alternatives [13], job selection, optimal maintenance strategy selection, selection of online taxi companies, and others. Bowles and Peláez [14] introduced a new technique based on fuzzy logic for prioritizing failures for corrective actions within Failure Mode, Effects, and Criticality Analysis (FMECA). In their study, the authors represented Severity (S), Detectability (D), and Occurrence rate (O) as fuzzy sets to evaluate failure modes in FMECA. Subsequently, defuzzification is performed using the maximum weighted mean method to assess failure risk. Ravi Sankar and Prabhu [15] proposed a new Risk-Priority Ranking (RPR) technique that uses a rating scale from 1 to 1000 to represent increasing risk levels for combinations of S, O, and D parameters. They also argued that conventional FMEA attempts to determine risk without adequately quantifying the contributing factors; therefore, the RPN can be misleading. Pillay and Wang [16] introduced Evidential Reasoning (ER) using a fuzzy rule base and grey relational theory to compute the risk priority score, thereby overcoming shortcomings of the traditional FMEA approach. Seyed-Hosseini et al. [17] employed the DEMATEL decision-making method to reprioritize failure modes in FMEA based on severity (impact) and the direct and indirect relationships among them.

The integration of FMEA with the F-AHP has proven to be a powerful approach for addressing the complexities of supplier selection by combining systematic risk assessment with MCDM. This integrated method, particularly when dealing with vague or subjective criteria, enables more accurate evaluations. Empirical studies across industries such as automotive, electronics, and construction have confirmed the positive impact of this hybrid approach on the accuracy and transparency of decision-making processes. For example, Fattahi et al. [18] demonstrated that the use of FMEA–F-AHP can effectively reduce supply chain risks. Similarly, Altubaishe and Desai [19] emphasized that this method yields superior results compared to traditional techniques in dynamic and complex environments. The integration of FMEA with the F-AHP has become a cornerstone of supplier selection, effectively addressing complex MCDM challenges. This hybrid approach combines detailed risk analysis with subjective judgment, making it ideal for applications across various industries. Previous studies, such as Başaran and Ighagbon [20], have shown the applicability of this hybrid approach in complex evaluation scenarios, including risk assessment and quality evaluation of mobile learning platforms. Their findings highlight the unique advantage of integrating these methods to provide a robust decision-making framework. Moreover, the integration of FMEA and F-AHP has been successfully implemented in industrial settings. For instance, Parmar et al. [21] evaluated failure modes in hydraulic direct drive systems using a combination of F-AHP and TOPSIS, highlighting its application in supply chain risk management. This hybrid method has also found relevance in the healthcare sector. Golrizgashti and Keshmiri [22] applied an AHP–FMEA model to assess patient safety and emphasized the adaptability of these techniques across industries.

Chen [23] introduced a modified FMEA that evaluates the interdependencies among corrective actions using Interpretive Structural Modeling (ISM). In this research, the author calculated the weight of corrective actions through the Analytic Network Process (ANP) and successfully combined them with utility priority numbers to support decision-making on FMEA improvement priorities. Wang et al. [24] proposed Fuzzy Risk Priority Numbers (FRPNs) that account for the imprecision of human judgments. These FRPNs overcome the limitations of conventional RPNs. They are subsequently defuzzified using the centroid method, in which a new centroid-based fuzzification formula derived from α-level sets is employed. Sachdeva et al. [25] applied FMECA to maintenance problems and noted that FMECA does not account for interdependencies among failure modes and effects. To address this limitation, they employed Shannon entropy to assign objective weights to maintenance parameters and the TOPSIS method to determine the maintenance criticality index. Sellappan and Palanikumar [26] proposed a modified FMEA in which the risk-priority-number prioritization method is revised. Their study highlighted a major limitation of conventional RPNs: different combinations of severity, detectability, and occurrence can yield identical RPN values while leading to completely different risk consequences.

Arabsheybani et al. [27] proposed an integrated fuzzy MOORA–FMEA method for evaluating sustainable supplier risks, accounting for quantity discounts, and highlighted its suitability for multidimensional decision-making scenarios. An integrated Fault Tree Analysis (FTA) and F-AHP approach for green supply chain risk assessment demonstrated the capability of hybrid methods to enhance sustainability [28]. Yazdani et al. [29] examined supplier selection using DEMATEL, FMEA, and EDAS, highlighting its effectiveness in risk management and in strengthening strategic decision-making for environmentally focused projects. Tavana et al. [30] developed an integrated AHP and MULTIMOORA framework that advanced the analysis of supply chain risks and benefits, particularly for supplier prioritization. In another study, researchers integrated AHP–PROMETHEE with FMEA for supply chain management, expanding its application within hybrid decision-making frameworks [19].

Growing interest in this approach is evident across various studies, with researchers supporting its application in diverse domains, ranging from rail transportation systems [31] to optimization of digital twin systems [32]. Despite its widespread application, certain challenges remain. As noted by Gul [33], researchers need to refine these integrated methods to address evolving decision-making scenarios, including adapting datasets and increasingly complex decision hierarchies. In supplier selection criteria, supplier risk depends on the type and magnitude of risk. Supplier-induced failures pose a hazard to manufacturers, who strive to assess and score each failure's impact. The aggregate score reflects the level of supplier risk; thus, the preferred supplier selection method is effectively equivalent to selecting the supplier with the lowest risk. Although many researchers have studied supplier evaluation and selection problems, only a limited number have examined them in terms of supplier or supply chain risk. In any decision-making problem, not all criteria have equal influence, and decision criteria play a critical role in the decision system. However, a review of the literature reveals that most researchers focus on efficiently prioritizing conventional RPNs in FMEA, while only a few concentrate on the relative weights of decision criteria. Integrated FMEA can overcome this limitation. Therefore, the present study focuses on the supplier evaluation problem using an integrated FMEA approach. Consequently, the integrated FMEA–F-AHP framework can be regarded as a versatile tool for supplier selection, offering a systematic and consistent approach to MCDM. The potential for future advancements lies in integrating advanced algorithms and expanding applications within dynamic supply chain environments.

In the supplier selection process, the decision-making criteria relevant to the system must be identified and considered first. Subsequently, potential failure modes should be recognized. The effective implementation steps for this research are presented in a flowchart (Fig. 1). Briefly, based on expert opinions and previous studies, the main criteria and sub-criteria for supplier evaluation are initially selected. In the next stage, the weights of the criteria must be assessed. For this purpose, the F-AHP is employed to minimize the influence of human judgment. To account for risk and uncertainty in the problem, the integrated FMEA technique is used as a key step in this study. Accordingly, the RPN is calculated based on experts' opinions and combined with the F-AHP-derived weights, yielding the final weights for criteria and sub-criteria for different suppliers. Based on these calculations, the most desirable supplier is identified according to the

Sobhan Darou company (Public Joint Stock) was originally designed and established in 1976 under the name Sooiran Cosmetics and Hygiene company (Private Joint Stock). The factory was launched in 1981 and renamed Sobhan Pharmaceutical company in 1985. Following the establishment of the factory and new production lines in 2004, it was registered as Sobhan Darou company (Private Joint Stock). In 2005, the company changed from a private joint-stock to a public joint-stock company. In 2010, its name was listed on the Securities and Exchange Organization under the symbol "DSobhan".

The expert team consists of marketing, sales, and commercial managers; industry specialists with several years of experience in the pharmaceutical sector; quality assurance experts; and academic members, all of whom are tasked with validating the criteria extracted by the research team based on previous studies. The industrial/academic experts were asked to rank the main drivers of the integrated FMEA model by their impact on various selection criteria for the best supplier, including social, sustainable quality, and economic factors.

Generally, economic criteria are defined in terms of costs and profits. In the production context, the economic aspect of sustainability is described in terms of its impact on shareholder economic welfare and local and national economic strategies [34]. The economic dimension covers all economic indicators, including conventional criteria related to financial accounting and intangible assets that are not typically reflected in financial reports [34]. However, the economic aspect is inherently aligned with both social aspects and sustainable quality. The social and sustainable quality dimension is based on empowerment or access to resources, employment generation, loyalty, transparency, and formal stability. It considers community cohesion and its capacity to work toward mutual goals while meeting individual needs such as worker health and safety, nutrition, shelter, education, and cultural expression. In this study, three main suppliers of the organization were evaluated.

A hybrid approach was adopted to identify relevant criteria and key drivers, drawing on a literature review and expert opinions. As the first step, studies by Kirytopoulos et al. [35] and Ishizaka et al. [36] were used as a source to select 14 criteria for triple sourcing in terms of economic, sustainable quality, and social criteria in the raw materials supply sector. In addition, 42 key sub-criteria were selected from the most important sub-criteria identified in previous sources for the integrated top-supplier model. Subsequently, 15 experts were contacted via email, phone, and in-person interviews; 12 (80%) agreed to respond. The experts were then asked (via a designed questionnaire) to confirm and prioritize the 14 selected criteria and 42 key sub-criteria regarding the supplier selection approach and overall impact on sustainability. The response time frame was from October 3, 2024 (questionnaires sent) to November 8, 2024 (responses received).

In some cases, in-person and telephone discussions were conducted for clarification. After receiving all responses, the researchers analyzed the questionnaires. They ultimately selected eight criteria and 22 sub-criteria as the basis for evaluating suppliers using the integrated FMEA and fuzzy hierarchical analysis method.

In this study, the authors employed an integrated FMEA approach to select the best supplier in a supply chain risk environment. When RPNs are identical across different decision-making criteria, traditional FMEA cannot handle the situation. However, this integrated FMEA approach successfully overcomes the limitations of conventional FMEA. In this research, for confidentiality reasons, the exact names of the suppliers are not disclosed. None of the respondents raised any negative claims regarding the pairwise comparisons or risk numbers, and they were highly satisfied with the selection process. Despite the contributions, this study still has some limitations. In the F-AHP method, triangular fuzzy numbers are used to account for human judgment imprecision. To further improve judgment accuracy, trapezoidal fuzzy numbers could be utilized. This study is based on information provided by the suppliers, and the relative weight of each criterion may change when applied to a different organization. In this research, eight criteria and twenty-two sub-criteria were considered, but they may not be sufficient for other organizations. Besides F-AHP, other MCDM methods, such as Fuzzy Analytic Network Process (F-ANP) and Fuzzy TOPSIS, can also be integrated with the FMEA technique. The authors also suggest using FTA, the Delphi method, or stress testing in addition to FMEA and combining them with MCDM approaches