State estimation and sensor location for Entrained-Flow Gasification Systems using Kalman Filter

Abstract

This study presents the estimation of key unknown states for a pilot-scale gasifier using Kalman Filter (KF). The transient behaviour of a pilot-scale gasification unit is represented using a dynamic reduced order model. This model consists of 479 state variables including molar fractions for the species, temperature, and slag thickness across the gasifier. The quality of the state estimation provided by KF has been evaluated under multiple arrangements of the number and the location of the sensors available for the top section of the gasifier. Also, plant-model mismatch, additive uncertainty in the prior estimation, and load-following scenarios have been considered. The results show that KF is capable of estimating the unknown states for a large variety of changes in the gasifier’s inputs, even though online temperature sensors are only available in limited locations across the gasifier.