Considerable attention has been paid to the heat-integrated water network synthesis (HIWNS) because it has the advantages of reducing water consumption, energy consumption and total cost. In this work, a revised superstructure of Ahmetovic and Kravanja (2013a Energy 57, 236-250) for simultaneous HIWNS is developed by changing the position of the heaters and coolers. Based on the superstructure, a nonlinear programming (NLP) model is developed for the HIWNS. We develop a method to denote the existence of a process match, which is generally addressed using discrete variables. In addition, the temperature difference terms are incorporated into the objective function using a reformed approximated equation for the logarithmic mean temperature difference (LMTD); therefore, the temperature approach variables are not necessary in the NLP model. A method to identify the stream roles as hot or cold streams is proposed with no discrete variables. The number and type of variables are reduced by these strategies. Branch-And-Reduce Optimization Navigator (BARON) solver is used to solve the NLP model with the default initial point (the lower bounds) by GAMS. The model was tested on single- and multiple-contaminant problems using seven cases, where one case has an identical total annual cost (TAC) to the literature value and the other six cases have smaller TAC than the literature reported values. (C) 2015 Elsevier Ltd. All rights reserved.