A unified approach to multiscenario sensor network design for data reconciliation

The design of sensor networks for chemical plants is an important and complex task. In the context of data reconciliation, there are several desiderata for a sensor network, such as observability, redundancy (including hardware redundancy), and estimation precision, as well as good performance in the process optimization conducted after data reconciliation. While each of these aspects has been considered in the literature, there is no published method that addresses all of them simultaneously. In this work, we develop such a unified approach while also considering the performance of the selected sensor network across multiple operating scenarios. We formulate the problem as a mixed-integer nonlinear program and highlight the benefits of the proposed integrated model in three computational case studies. In particular, the results show that hardware redundancy can lead to improved performance at the same sensor cost and that a multiscenario approach can be crucial in achieving overall optimality.