In the competitive automotive parts manufacturing industry, identifying and prioritizing failure modes is essential for preventing financial losses, ensuring passenger safety, and maintaining quality. This applied, descriptive-analytical study aimed to identify and prioritize failure modes in the constant-velocity (CV) joint assembly process of an Iranian automotive parts manufacturer, integrating the Fuzzy Best-Worst Method (FBWM) and Fuzzy VIKOR (FVIKOR). Fifteen purposively selected experts participated in a three-round Delphi survey to screen risk factors and failure modes; retained factors were weighted using FBWM and failure modes ranked using FVIKOR. Severity received the highest weight among retained risk factors, and failure modes associated with tensile strength and dimensional and spline tolerances of CV-joint components received the highest corrective-action priority. Expert pairwise comparisons showed high consistency, and combining Delphi screening with fuzzy multi-criteria weighting and ranking improved prioritization accuracy compared with the traditional risk priority number approach, providing production managers with a reliable basis for preventive decision-making.
Introduction
The automotive parts manufacturing sector in Iran faces persistent challenges related to product quality, safety, and reliability. Failure Mode and Effects Analysis (FMEA) is one of the most widely used preventive quality management tools, designed to identify and prioritize potential failures before they occur (McDermott et al., 2009; Baghery et al., 2018). The traditional Risk Priority Number (RPN)—the product of severity, occurrence, and detection—has well-documented limitations: identical RPN values can arise from different risk combinations, equal weighting is assumed, and crisp scores cannot capture expert judgment ambiguity (Liu et al., 2013; Yousefi et al., 2018). Recent literature called for replacing RPN with fuzzy multi-criteria approaches (Liu, 2016; Akkus, 2024; Liou et al., 2024). This study integrated three complementary methods (Delphi screening, FBWM for criterion weighting, and FVIKOR for failure mode ranking) in the context of CV-joint assembly, a safety-critical powertrain component. No prior study had simultaneously applied all three elements in the Iranian automotive parts sector, nor provided empirical sensitivity validation and an RPN comparison.
Methodology
This applied, descriptive-analytical study was conducted via a survey approach. A panel of fifteen purposively selected experts from a CV-joint manufacturer—drawn from quality, production engineering, R&D, sales, and senior assembly roles, each with at least ten years of experience—participated in all study phases. Delphi screening used a five-point Likert scale with an acceptance threshold of 3; termination required a Kendall’s W change of less than 0.02 between rounds and no further item elimination. Five candidate criteria (severity, occurrence, detection, cost, time) were screened in three rounds. The four retained criteria were weighted using FBWM with the nonlinear optimization model solved in Lingo; weights were defuzzified using the Graded Mean Integration Representation (GMIR) method (Chen & Hsieh, 1999) as standardized in Guo and Zhao (2017). The consistency ratio was computed as CR = k/CI. Seventy-one potential failure modes were reduced to 68 through three Delphi rounds. The 68 retained failure modes were evaluated using fuzzy linguistic scales (Patil & Kant, 2014) and ranked by FVIKOR (Opricovic, 2011) with compromise parameter v = 0.5, defuzzifying S, R, and Qusing (l + 2m + u) / 4. A sensitivity analysis across v ∈ {0, 0.25, 0.5, 0.75, 1} and an illustrative RPN comparison were also conducted for a 12-FM subsample.
Findings
FBWM yielded weights of 0.408 (severity), 0.265 (cost), 0.186 (occurrence), and 0.139 (detection). The consistency ratio CR ≈ 0.064 confirmed high expert agreement. The “time” criterion was eliminated in Delphi round 1 (mean = 2.933). Three failure modes were eliminated in the failure-mode Delphi round 1. Cronbach’s alpha was 0.91 (round 2) and 0.90 (round 3); Kendall’s W was 0.247 (round 2) and 0.254 (round 3). FVIKOR ranked all 68 failure modes; the five highest-priority were: FM32—Insufficient tensile force in the trunnion test (Q = 0.0195); FM65—Low stopper bearing tensile strength (Q = 0.0349); FM56—Spline deformation (Q = 0.0352); FM4—Dimensional non-conformance (Q = 0.0364); FM13—Shaft contact with boot edge (Q = 0.0468). The RPN comparison showed that failure modes with high detection difficulty but low severity (FM64, FM67) were over-ranked by the simple additive approach relative to FVIKOR. The sensitivity analysis demonstrated full robustness: {FM3, FM4, FM65, FM66} occupied the top four positions across all five v values, and the ordering of the remaining eight subsample failure modes was completely unchanged.
Discussion and Results
The findings confirmed that integrating Delphi screening, FBWM weighting, and FVIKOR ranking provided a robust approach to failure mode prioritization. The dominance of severity (weight = 0.408) reflected the safety-critical nature of the CV joint. FVIKOR’s superiority over simple additive scoring was most apparent when criticality was concentrated in high-severity or high-cost criteria. The study extended existing literature as the first to simultaneously apply all three elements (Delphi, FBWM, FVIKOR) in Iranian automotive parts manufacturing, and provided empirical sensitivity validation and an RPN comparison. A key limitation was that raw decision matrix data could not be fully disclosed due to company confidentiality.
Conclusion
This study demonstrated that the Delphi–FBWM–FVIKOR framework effectively identified and prioritized failure modes in CV-joint manufacturing. The three highest-priority failure modes—insufficient tensile force in the trunnion test, low stopper bearing tensile strength, and spline deformation—warrant immediate corrective action. The consistency (CR ≈ 0.064) and robustness (full rank stability across all v values) confirmed the reliability of the approach. Future research should replicate the framework across multiple production lines and explore K-means expert clustering prior to FBWM weighting, following Amoozad Mahdiraji et al. (2022).