A novel failure mode and effect analysis model for machine tool risk analysis

Highlights

• We integrate several state-of-the-art methods to analyze system/product failures.

• The proposed model combines the rough number, BWM, and modified TOPSIS techniques.

• The variety of expert opinions can be effectively integrated with this method.

• The proposed model is demonstrated using data provided by a machine tool company.

• Comparisons are performed between results from the proposed model and past methods.

Abstract

Increasing the reliability of machine tools and reducing possible risks during the manufacturing process is crucial for the future of industry. The failure mode and effects analysis (FMEA) method is reliant upon the experience of experts to determine the primary failure modes and detect the most critical factors for preventing risk. Clearly, an effective method capable of integrating the various different expert opinions is required. This study proposes a novel FMEA model based on multi-criteria group decision-making, which is developed by integrating a rough best–worst method, and modified rough technique for order preference by similarity to an ideal solution for ranking failure modes. The model can overcome some of the limitations of the conventional FMEA. It also includes the expected cost as a risk element to provide a more practical result. The effectiveness of the proposed model is demonstrated by conducting a case study involving a machine tool company. The results indicate that the proposed model can effectively assist managers in evaluating risk factors and identifying critical failure modes.

Introduction

Machining processes are critical production procedures in modern manufacturing processes that require the use of equipment ranging from lightweight mobile devices to heavy machinery [1]. Continual improvement in production quality and efficiency, the development of machine tools with greater reliability and accuracy, is the trend [2]. In general, if machine tools do not operate appropriately, production must be immediately discontinued. The service engineer must assists the machine tool customers to diagnose the errors and resolve the problem. However, the best solution would be predetermination of probable failure modes, which would assist manufacturers in designing preventive or corrective measures, which in turn could reduce failure repair time and probability of occurrence, thereby enhancing reliability [3]. The so-called “intelligent machine tools” have recently received considerable attention from manufacturers but with their adoption, troubleshooting and maintenance are crucial parts of the monitoring process and indispensable for machine tool users.

In past studies, quantitative approaches for system failure identification have been scarce and often applied only for cost evaluation, as described by Dehghanian et al. [4]. Reliability centered maintenance (RCM) is a management methodology for maintenance planning that emphasizes quantitative analysis by means of mathematical logic, especially for equipment intensive industries [5], [6]. The application of the RCM methodology has been based on traditional approaches, such as failure mode and effect analysis (FMEA), fault tree analysis (FTA), hazard and operability studies, and diagrams and algorithms [7]. FMEA has been the most commonly used and effective method for the development of fault diagnosis and risk assessment models in recent years. Therefore, this study proposes a quantitative analysis method based on FMEA for identifying critical failure modes and improving the robustness of machine tools.

The FMEA technique is frequently employed for detecting complex system failure modes for the eradication of risk factors and to increase product reliability and safety [8]. However, there are limitations to the conventional FMEA methods including the following:

• Simply multiplying the severity (S), occurrence (O) and detection (D) ratings to find the risk priority number (RPN) is too simplistic and strongly sensitive to variations in the RPN element evaluations [9];

• S, O, and D are weighted equally, but because risk analysis cases differ, their weights should not be equal [10];

• Only the three RPN elements S, O, and D are considered in the analysis [11];

• Different S, O, and D evaluation methods may provide the same RPN [12].

These limitations can be overcome by combining multi-criteria decision-making (MCDM) methods with the FMEA [8,13]. Qualitative methods for evaluating the weights of the RPN elements include the analytic hierarchy process (AHP) and analytic network process (ANP). MCDM also provides several methods for prioritizing failure modes. Examples of such methods include the technique for order preference by similarity to an ideal solution (TOPSIS), decision-making trial and evaluation laboratory (DEMATEL), complex proportional assessment (COPRAS), and grey relation analysis (GRA). All these methods have been applied to analyze the RPN and have provided a basis for risk assessment [14], [15], [16], [17], [18].

Rezaei [19], [20] proposed a new comparison-based method, a best–worst method (BWM), for calculating the weights of criteria. The BWM performs comparisons in a structured manner that requires fewer pairwise comparisons than the AHP and makes it easier to achieve results. Moreover, Kuo [21] developed a new ranking index, an improvement over the original TOPSIS that not only considers the distances of all alternatives to a positive ideal solution and negative ideal solution but also considers the weights of these ideal solutions. The ranking index was examined in several experimental tests, and the results demonstrated that this modified TOPSIS is highly reliable in practice. There has been no FMEA which has used the BWM to obtain the RPN element weights and applied the modified TOPSIS to sort the failure modes.

In this new method the risk value assessments are provided by experts as input for FMEA analysis. However, the experts may not reach a consensus regarding how to remediate or avoid the failure. It is necessary to aggregate the opinions of various experts with different backgrounds. In this study, expert opinions were synthesized using a rough number-based approach, which is effective for multi-criteria group decision-making (MCGDM). The rough number approach allows the aggregation of individual observations and preferences to be used in the decision-making process thereby overcoming the shortcomings of using the arithmetic mean method. With this method no statistical assumptions are required, and the conceptual logic is intuitive and simple. The contributions and improvements of the proposed model are outlined below.

• The number of RPN elements can be increased to make the assessment more comprehensive. The proposed model includes the expected cost (E) in the RPN calculation.

• Weight calculation is more efficient and simpler. The BWM requires fewer pairwise comparisons and there is better consistency.

• The proposed R-BWM and R-TOPSIS methods can effectively overcome the shortcomings of methods based on the arithmetic mean and avoid the loss of valuable expert information.

• The method is flexible and reliable for practical applications. The input data is based on qualitative information from the expert surveys, which is different from the quantitative data required for fault tree analysis.

In summary, the proposed model concept is novel. The ranking results of the failure modes can accurately represent real-world situations and thus assist engineers to formulate accurate improvement guidelines, thereby enhancing the robustness and lifecycle of machine tools.