Bayesian methods are suitable for industrial sites, where there is a good understanding "a priori" of deterioration, failure and rupture mechanisms. This understanding comes form decades of cumulated experience, which is the basis of knoweledge. If Knowledge, instead, is poor, may be the case of a very new industry, bayesian approach could be misleading ....
Of course the only disadvantage is the choice of priors. The priors that one selects may be subjective and may lead to unreliable prediction and possibly misleading results. Unless such priors are indeed objective, then a Bayesian method is absolutely not a great way to follow for predicting reliability.
I agree with John about the Bayesian method. Also, Continuous Time Markov Chain (CTMC) can be an alternative choice without having the issue of dealing with priors. However, CTMC has other disadvantages like state explosion and the limitation of choosing the failure distribution function. For the first one there are some rules to simplify the model, but still, the problem exists. For the second one, it is possible to use more advanced methods such as Semi-Markov Process, Regenerative Markov Process, and Phased Type Models.
Modelling a system with Weibull failure distribution can be an example for the second mentioned disadvantage than can be considered through advanced methods.
Continuous Time Bayesian Network (CTBN) allows to model root nodes (priors) as continuous variables. However, with continuous variables, it may become tedious to express the joint probability distribution of internal nodes with many parents with a probability density function.
In addition to the issue with priors, if the parent-child relationships (e.g., CPTs) among the nodes are not accurately defined, then we will get an inaccurate (possibly misleading) estimate of system reliability. If the parent-child relationships are not deterministic (e.g., BN models of logic gates have deterministic relationships between parent and children), then significant knowledge about the system is needed to define these relationships accurately.
Bayesian methods are suitable for industrial sites, where there is a good understanding "a priori" of deterioration, failure and rupture mechanisms. This understanding comes form decades of cumulated experience, which is the basis of knoweledge. If Knowledge, instead, is poor, may be the case of a very new industry, bayesian approach could be misleading ....
But additionally, during constructing of the BN graph, you have to pay much attention to the common cause failures and conditional dependency in order to represent the actual system behavior.