DYNAMICS AND CONTROL OF FORCED UNSTEADY-STATE CATALYTIC REACTORS

This research deals with the dynamics and control of forced unsteady-state catalytic reactors and it is focused on two topics: 1. auto-thermal after-treatment of lean VOC mixtures. Two reactor configurations have been taken into consideration: the reverse-flow reactor (RFR), where the flow direction is periodically changed, and the network of two or three reactors (RN), where the flow direction remains the same, but the feeding position is periodically changed, thus simulating a moving bed. This study (§3) has been organised as follows: - modelling of the two reactor configurations and study of the influence of the main operating parameters (§3.1 and §3.2). As the RFR shows higher stability with respect to disturbances in the feed a deeper investigation has been carried out on this device; - optimisation of the RFR. A simplified model has been used for this analysis in order to strongly reduce the computational effort which is required by detailed models. It has been pointed out that both heat capacity and thermal conductivity of the catalyst play a role, not less important than kinetic activity, strongly influencing the minimum inlet VOC concentration required for autothermal operation (§3.3); - experimental validation of the modelling results in a bench-scale RFR with reduced influence of the wall effects. This activity has been carried at the Departamento de Ingeniería Química y Tecnología del Medio Ambiente-Universidad de Oviedo (Spain) in the framework of the Research Project “Azioni Integrate Italia-Spagna”, granted by the Italian Ministry of Research (MIUR). In addition to the intrinsecally dynamic behaviour of the RFR, one must deal with unexpected external perturbations (feed concentration, composition and temperature) which may lead to reactor extinction or catalyst overheating. In order to avoid these problems it is necessary to implement some closed-loop control strategy based on the measurement of the inlet concentration (and composition) and the outlet conversion. This study has been organised as follows: - a model-based soft-sensor (observer) has been developed, in order to quickly and reliably estimate the feed composition from some temperature measurements in the reactor, thus avoiding expensive hardware sensors and time consuming on-line measurements. As deriving an observer from a detailed model is an overwhelming task, a simplified model has been developed and validated in a medium size RFR. This research has been carried out in cooperation with prof. H. Hammoury and D. Schweich of the CPE-Lyon, France (§4.1); - a Model Based control strategy has been proposed and tested to prevent reaction extinction and catalyst overheating (§4.2); 2. enhancement of conversion and selectivity in exothermic, equilibriumlimited reactions. Methanol synthesis and syngas prouction by partial oxidation of methane have been considered as test reactions. This section has been organised as follows: - modelling of the two processes in the two reactor configurations previously described. The influence of the main operating conditions has been addressed with the aim to optimise the two processes. As the RN has shown higher conversion and selectivity with respect to the RFR, in the following the research will be focused on this device (§5); - a simple open loop control policy, which can be useful for a safe startup, has been also tested to study the response of the RN to disturbances on the input parameters, showing that a more robust control strategy is needed for this application; - if a tight control on the outlet product conversion is needed, a Model Predictive Control scheme (MPC) should be used, varying the switching time to maximise the conversion and the selectivity of the reactor. The on-line optimisation requires a simplified model and a Neural Network based model has been developed (§6).