Impact of plant dynamics on the performance of steady-state data reconciliation

This paper studies the impact of plant dynamics on the performance of steady-state data reconciliation for processes operating in stationary mode. Data reconciliation can effectively reduce effects of measurement errors on process variables estimation. This has greatly motivated the use of steady-state observers in the process industry. However, processes are operating under unsteady-state conditions and the fundamental hypothesis of steady-state is not respected. The objective of the paper is to develop criteria for assessing the impact of stationary disturbances on estimates quality and therefore to propose applicability indices for steady-state data reconciliation methods in disturbed plants. A generic criterion based on generalized variance is given and criteria are proposed for plant performance indicators using variance reduction in specific state-space directions. The various applicability indices are illustrated through a material balance example for a simulated mineral separation circuit and the steady-state approach is compared to an optimal stationary data reconciliation method.