The manufacture of nonionic surfactants generally involves ethoxylation via ethylene oxide condensation onto a hydrophobe substrate, mostly in the presence of an alkaline catalyst. Nonionic surfactants are used widely in industrial applications, such as detergents, health and personal care, coatings, and polymers. In Australia, approximately one-third of the annual consumption of nonionic surfactants is imported from offshore manufacturers; the market is highly competitive with the local manufacturer facing increasing competition from imports. Optimisation is a pressing need for the current manufacturing plant of the industrial partner for this research project, Huntsman Corporation Australia Pty Limited, the sole domestic manufacturer of nonionic surfactants in Australia. Therefore, the objectives of this research project were to gain a better understanding of the various chemical and physical processes occurring simultaneously in an ethoxylation process; to identify the process limitation in an existing production plant operated by Huntsman Corporation Australia, and to explore measures for enhancing the asset productivity of the production plant. An ethoxylation process working model, describing the chemical kinetics and the physical transport processes involved, was developed to aid the exploration of optimisation opportunities, which would otherwise be empirical. Accordingly, this research project was structured into a two-stage program. The first stage determined the ethoxylation kinetics experimentally. The second stage investigated the interactions of physical transport processes numerically using a computational fluid dynamics (CFD) technique. The manufacturing scheme discussed in this thesis gave particular emphasis to the ethoxylation process operated in semi-batch stirred reactors. In the first stage, a series of kinetic experiments was performed in a well-stirred laboratory autoclave under base-catalysed conditions. The experimental outcomes were developed into a comprehensive kinetic model which took into account the non-ideal features in the reactor operation. Time-dependent physical changes of the reaction system, such as liquid volume, ethylene oxide solubility and density were also included. The ethoxylation behaviour predicted by the model was shown to be in good agreement with the experimental measurements. This indicated that the kinetic model was sufficiently robust to reproduce the reaction behaviour of a commercially operated ethoxylation operation. In the second stage, numerical simulations of an existing ethoxylation reactor system were presented. In addition, two components were addressed: identification of the process limitation and increasing productivity of the industrial-scale ethoxylation plant. An important assumption was made for the ethylene oxide injection system used in this research project which subsequently simplified the ethoxylation system into a single liquid with miscible chemical species. In the identification of the process limitation, three possible rate-limiting factors were examined: mixing, heat removal and reactor pressure rating. Examination and analysis of the physical data available from plant batch reports found that the reactor pressure rating and the presence of nitrogen padding were the rate-limiting factors to the ethoxylation operations in the industrial reactors. It was recommended that the reactor pressure rating be increased to raise the asset productivity of the reactor. In the numerical simulations of the ethoxylation reactor, time-dependent CFD models were developed for two systems: the ethylene oxide injection pipe and the stirred ethoxylation reactors. The heat transfer of ethylene oxide liquid injection was calculated in a two-dimensional model of the dip-leg pipe used in an industrial-scale ethoxylation reactor. The computation gave the temperature of the injection outflow which was validated against the calculated value by empirical correlation. The effects of various surrounding reaction temperatures, injection rates and pipe sizes on the heat transfer rate were investigated. From these, a range of operating conditions yielding a liquid ethylene oxide outflow was selected. Furthermore, it was found that boiling of ethylene oxide was significantly reduced with increasing pipe diameters. It was recommended that the asset productivity of the reactor be improved by keeping more ethylene oxide injected as a liquid in the reaction mixture to raise the reaction rate and shorten the reaction time. Three-dimensional simulations of a baffled reactor agitated by a single- or a dual-Rushton impeller were presented for both non-reactive and reactive flows. Multiple frames of reference and sliding grid methods were used in sequence to describe the relative motion between the rotating impeller and the stationary baffles. The turbulence parameters were modelled with the standard k- � turbulence model. The simulations of non-reactive flow were compared with the literature velocity data obtained from both the experiments and simulations. Good agreement was achieved. The model was then extended to incorporate ethoxylation flow with integration of the kinetics established in the first stage. Both the laboratory autoclave and the industrial-scale reactors were simulated. The former took into account the ethoxylation exotherm and the latter was carried out isothermally. Both simulations were validated against reaction data obtained from physical experiments, either the kinetic experiments or the plant batch productions. The validated model allowed us to determine the optimum operating condition and explore a new reactor system with enhanced asset productivity. A 50% increase in productivity could be accomplished if the ethoxylation was operated closer to the current design pressure limit. Furthermore, the operating pressure of a new reactor system needed to be doubled if the asset productivity were to be increased to approximately three times the current performance.

Thesis (PhD) - Swinburne University of Technology, Faculty of Engineering and Industrial Sciences, Centre for Micro-Photonics, 2005. / A thesis submitted in fulfillment of the requirements for the degree of Doctor of Philosophy, Centre for Micro-Photonics, Faculty of Engineering and Industrial Sciences, Swinburne University of Technology, 2005. Typescript. Bibliography: p. 201-210. Also available on cd-rom.