This paper describes the development of a decision support system (DSS) for evaluating membrane restoration strategy. The engine of the DSS is a digital twin (DT), a virtual representation of wear (degradation) and restoration of membrane elements in a reverse osmosis (RO) pressure vessel. The basis of the DT is a mathematical model that describes an RO vessel as a novel multi-component system in which the wear-states of individual elements (components) are quantified and elements can be swapped or replaced. This contrasts with the contemporary presentation in the literature of a membrane system as a single system. We estimate the parameters of the model using statistical methods. We describe our approach in the context of a case study on the Carlsbad Desalination Plant in California, which suffers from biofouling due to seasonal algae blooms. Our results show a good fit between the observed and the modelled wear-states. Competing policies are compared based on risk, cost, downtime, and number of stoppages. Projections indicate that a significant cost-saving can be achieved while not compromising the integrity of plant.