Article Preview
TopIntroduction
The prioritization of equipment plays a crucial role in maintenance management, it allows steering the different maintenance sub-functions towards targeting the top priority equipment, ensuring it gets suitable maintenance strategies and enough spare parts along with the allocation of an adequate budget for its specific maintenance activities. Thus, allowing to attain a stable production, respecting the quality standards and the delivery dates. Considering that the equipment priorities are defined according to multiple criticality factors; often conflicting, the Multi-Criteria Decision Making (MCDM) framework can offer optimal solutions in this context, allowing to improve the overall maintenance performance while ensuring the profitability of its actions, simultaneously. There are some well-known MCDM techniques that have been successfully implemented for solving maintenance prioritization problems such as TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution), PROMETHEE (Preference Ranking Organization METHod for Enrichment Evaluation), ANP (Analytic Network Process) and AHP (Analytic hierarchy process). Among these, the latter seems to be often preferred; in a recent review on maintenance prioritization approaches, Chong et al. (2019) reported that AHP is by far the most frequently used method. This might be due to various reasons; starting by its ability to decompose a complex problem into a simplified hierarchy of multiple levels. Not to mention that it is among the few methods which embed a weighting technique, allowing to generate reasonable weights reflecting the real importance of the criteria, rather than specifying them explicitly; which may leave room for imprecise estimates as the criteria number increases. It even provides a mechanism to verify the inconsistencies present in the decision-maker judgements, in order to ensure a robust model design. Overall, it can process MCDM problems, efficiently. But apparently, this comes at the expense of time and effort consumption for its implementation, as AHP require 
pairwise comparisons, for a given number 
of criteria, while there are new methods in the literature that provide consistent judgements with way less comparisons, such as the Best Worst Method (BWM) and the Full Consistency Method (FUCOM), which require only 
and 
comparisons, respectively. Therefore, Saaty (the developer of AHP) recommended to use it with no more than 7 criteria, while other researchers advise to apply it only for a limited set of alternatives, for the same reason. Indeed, there are much bigger concerns associated with AHP; such as the well-known rank reversal problem and the conflicting results issue (Bafahm & Sun, 2019), reported recently. It is argued that one of the main reasons behind these problems is that the conventional AHP overlook the actual performance of the alternatives, as it relies on the relative performance comparisons only. Researchers are still trying to prevent or at least justify these problems, but none is settled yet. Furthermore, it is known that the AHP method exists in various versions (Crisp, Fuzzy, Rough) and supports 2 synthesis modes (Distributive, Ideal), which might confuse the decision maker willing to use it, given that the results of different versions of the same technique are mostly not even close. In this context, the MCDM techniques have been criticized for yielding up differentiated results when applied to the same problem (Pourjavad & Shirouyehzad, 2011), which led to the emergence of several comparative studies, in response to this confusion; Popovic et al. (2018) compared the AHP method with the Conjoint analysis, where they concluded that the latter is preferred for relatively simple decisions, while AHP is more convenient for the opposite cases. Dehghanian et al. (2011) showed that the fuzzy AHP is more efficient than its crisp counterpart in dealing with uncertainties, while Stanković et al. (2019) proved that this property is even further improved in Rough AHP by excluding the subjectivism. On the other hand, the comparison between the distributive and ideal synthesis modes has never been addressed in applied Studies. According to Millet and Saaty (2000), the distributive mode performs a ranking according to the dominance, while the ideal mode proposes a ranking according to the performance. So, the ideal mode is hypothetically more adapted to the equipment prioritization problem, given that it is basically a performance-based ranking problematic. However, this assumption has never been approved through an empirical study. Indeed, it is noticed that even the relevant applications tend to the use of the distributive mode rather than the ideal mode, which shows a clear conflict between the theoretical and practical aspects. Therefore, the goal of the current article is demonstrating that the ideal synthesis mode is more adapted to this problem, through a comparative study. This involves the development of an enhanced AHP-approach, bringing an improved accuracy and rightness, along with the support of a large number of equipment by overcoming the inconvenient pairwise comparisons issue. Furthermore, the proposed approach contributes in dealing with the issues associated with AHP mentioned previously, by incorporating the actual performance of the alternatives also into the model.