A rigorous method based on process equations for designing sensor networks for nonlinear systems was recently proposed (Nguyen and Bagajewicz, Design of Nonlinear Sensor Networks for Process Plants, Ind, Eng. Chem. Res. 2008, 47, 5529-5542). The method has been shown to be efficient for middle scale problems. For realistic large scale problems, the method can identify several suboptimal solutions but fails to identify the optimal solution within an acceptable PC computational time. This paper presents an approximate method that attempts to locate optimal solution starting from the suboptimal solutions provided by the rigorous equations-based method. We show that although optimality is not guaranteed, the proposed method was able to identify global optimum solutions in design case studies.