Mathematical Modelling of a Tubular Reactor for Continuous Production of Synthetic Resins

Abstract

Synthetic resins have, so far, been universally produced by batch processes, A continuous process however is shown to have some general advantages which make it desirable in commercial production. The physical properties of the resin systems have resulted in the choice of a tubular reactor for continuous operation and the urea-formaldehyde resin system is selected for detailed examination. To account theoretically for the performance of the tubular reactor two mathematical models are developed, The first model, which ignores all but the bulk transports, is a plug-flow model catering for good radial mixing in the reactor, The second model, involving a parabolic velocity profile with inclusion of radial and axial diffusion is a complex model and caters for laminar flow in the reactor tubes. The physical data necessary for the solution of the models are estimated using standard procedures. The chemical kinetic data available are shown to be inadequate for use at the elevated temperatures employed in the reactor. A novel technique is therefore developed for evaluation of the high temperature urea-formaldehyde reaction data, The data are then described mathematically with postulation of chemical reaction mechanisms, followed by optimisation of the rate constants for the proposed reaction schemes to give good mathematical fits. To test the models, an available rig is modified. The problems of sampling under reactor conditions are overcome by the introduction of a novel design of a sampling valve and sample collection technique which avoids physio-chemical change to the resin, The experimental results are discussed with regard to the performance of the reactor as affected by various parameters, The best conditions for production of the addition products of the urea-formaldehyde reaction with a minimum formation of the condensation products are shown to be high formaldehyde to urea molar ratios, high temperatures, low concentrations and low residence times. The plug-flow and complex models are solved using both a Honeywell 316 and an ICL 1904S computer. A novel numerical solution technique, in conjunction with the use of computers, is introduced for the solution of the complex models. The simulation results show excellent predictions by the plug-flow models, and poor accuracy of prediction by the complex models. The reason for the good plug-flow predictions, at comparatively low Reynolds numbers, is attributed to the shape of the reactor which also explains the shortcomings of the complex models.