The fundamental problem in optimal sensor network design is choosing a set of important or strategic process variables to be measured. An optimization formulation for sensor network design that relates process economics and data reconciliation is proposed. To address this, an economic quantity is defined to quantify the loss of operational profit caused due to measurement uncertainty. The resulting analytical expression that quantifies the loss is shown to be the sum of weighted error variances of the reconciled estimates obtained from reconciliation. The final formulation is a mixed integer cone program that can be solved to obtain a globally optimal sensor network. The effect of the process economics, capital cost, and marginal utility of additional sensors is illustrated using case studies.