As Water Distribution Networks (WDNs) age, they become increasingly susceptible to leakage, requiring comprehensive renovation strategies. While essential, rehabilitating these networks in dense urban areas faces strict environmental and operational constraints. Consequently, simultaneous network-wide renovation is often infeasible; instead, a sequential replacement of pipes is required to minimize disruption to city dynamics. However, this process inevitably disturbs urban functionality due to water supply interruptions and road network closures. Given the growing scarcity of global water resources, optimizing both water losses and delivery efficiency during this replacement sequence is critical. To address this, a Probability-Based Incremental Learning (PBIL) algorithm is employed. This thesis evaluates the algorithm's applicability and robustness across networks of varying sizes and configurations.