Water and energy are key resources in the process industry. The increasing pressure on freshwater and energy resources coupled with stringent environmental regulations on effluent discharge limits have called for innovative designs for sustainable use of water and energy. This can be achieved through process integration techniques that are environmentally benign and economically feasible. This work proposes a robust water network superstructure optimization approach for the synthesis of a multiregenerator network for simultaneous water and energy minimization. Two types of membrane regenerators are considered for this work, namely, electrodialysis and reverse osmosis. In each of the membrane regenerators, a detailed design model is developed and incorporated into the water network model. The detailed model presented in this work is compared to the more common "blackbox" model, which uses linear expressions to represent costs of regeneration units. The results show that the "blackbox" model gives inaccurate cost representation as compared to the detailed model and also the blackbox model framework does not give insights on optimal design of the regeneration units for minimum energy usage. The presence of continuous and integer variables, as well as nonlinear constraints renders the problem mixed integer nonlinear programming (MINLP). The developed model is applied to a pulp and paper case study to demonstrate its applicability, assuming a single contaminant scenario. The model was solved in GAMS using a solver BARON, and the results indicate a 43.7% freshwater reduction, 50.9% decrease in wastewater generation, and 46% savings in total annualized cost.