Measurements of process variables have a considerable impact on the control, optimization, safety, and reliability of chemical processes. In a 1993 article, Ali and Narasimhan developed a systematic graph-theoretic procedure for optimally locating a minimum number of sensors in linear steady-state processes. The sensor network was designed to maximize the probability of estimating variables when sensors are likely to fail. This article extends that procedure for the optimal design of a redundant sensor network for linear processes. The algorithm proposed also accounts for specifications of measurable and important variables. The efficiency and robustness of the proposed algorithm are demonstrated on realistically large processes.