The reliability of water supply system is a critical factor in the development and the ongoing capability to succeed in life and people's health. Determining of its, with high certainty, for performance of water supply system is developed to ensure the sustainability of system. Reliability (Re) plays a great role in evaluation of system sustainability. The probability approaches have been used to evaluate the reliability problems of systems. The probability approach is failed to address the problems of reliability evaluation that comes by subjectivity, human inputs and lack of history data. This research proposed two models; I) traditional model: fuzzy reliability measure suggested by Duckstein and Shresthaand then developed by El-Baroudy; and II) developed model: fuzzy reliability-vulnerability model. The two models implemented and evaluation of water supply system by using two hypothetical systems (G and H). System (G) consists of a single pump and System (H) consists of a two parallel pumps. Triangular and trapezoidal membership functions (MFs) are used to investigate of the reliability measure to the form of the membership function. The results agree with expectations that the reliability of parallel component system {ReH (0.53)} is higher than the reliability of single component system {ReG (0.47)}. Moreover, the result by using fuzzy set reduces the effect of subjectively in process of decision-making (DM). The fuzzy reliability vulnerability is able to handle different fuzzy representations and different operation environment of system
One way of obtaining information about reliability of units is to accelerate their life by testing at higher levels of stress (such as increasing elevated temperatures or voltages). Predicting the lifetime of a unit at normal operating conditions based on data collected at accelerated conditions is a common objective of these tests. Different models of accelerated life testing are used for such extrapolations. Two statistical based models are widely used: parametric models which require a prior specified lifetime distribution, and nonparametric models that relax of the assumption of the life time distribution. The proportional odds model is a nonparametric model in accelerated life testing based on the odds function and show that it gives a more accurate reliability estimates than proportional hazard model. This paper will concentrate on the models of proportional odds nonparametric accelerated life test for reliability prediction.