In this paper I focus on an evaluation of maintenance costs of a water distribution system (WDS), if a concept of a value of money in time is taken into account. Contrary to more classical approaches, instead of a constant yield, a strictly stochastic process (i.e., the one-factor Vasicek model) of an interest rate is assumed. Such an assumption presents uncertain, future behaviour of the yield in a more correct, realistic way. Moments of failures of connections in a WDS are generated using the Monte Carlo simulations via a new kind of a convex hazard rate function (HRF), which is proposed in this paper. Moreover, quality of a pipeline and a number of previous failures have direct influence on statistical properties of this introduced HRF. Apart from an analysis of the simulated output (like the maintenance costs), the Kiefer-Wolfowitz method is used for a better adjustment of one of parameters of a WDS – deterministic and unconditional replacement (i.e., planned replacement) time of each pipe. Algorithms, for both the simulations of the failure moments for the introduced HRF and the optimization step, are also provided. Additionally, some examples of a WDS for a crisp and a fuzzified settings are statistically analysed