Petrochemical manufacturing is a typical example of a continuous chemical process, which usually involves several interconnected process units. The unit interactions introduce many degrees of freedom for process operation, and also complicate supply chain management. Optimisation is therefore brought in to support decision-making on both process operation and supply chain aspects. In order to enhance operational feasibility and model accuracy, the optimisation should in principle be executed directly in a site-wide fashion and employing igorous process representations. This however results in an extremely large-scale site-wide model. This hinders the continuous process from obtaining the utmost benefit from the optimisation. This thesis hence aims to remove the obstacle by developing a practical direct optimisation method for site-wide supply chain problems. The thesis employed an ethylene process as an example of a continuous process for the solution method development. Based on the process, four types of problems with different complexities were established. They are (i) single-site single-period; (ii) single-site multi-period; (iii) multi-site single-period and (iv) multi-site multi-period. A novel direct optimisation technique, namely gPROMS-Based Application (gBA) Approach, was developed and examined using the example problems. It overcame the difficulty of the large problem size by separating the optiser from the process simulator, gPROMS. This allowed gPROMS to simulate each site and each period in the supply chain one-by-one, and send the required information to the optimiser individually. This avoided computer memory shortage problems, which could be triggered by a simultans optimisation ot the complete supply chain. Other solution approaches were also investigated in the thesis to compare with the gBA approach. For example, Successive Linear Programming (SLP) was used for the single-site single-period problem.

Chemical process synthesis is an open ended step of process design as it deals with the problem of how to develop and integrate the chemical process flowsheet. Over the past four decades, very few systematic procedures have been proposed for the rigorous synthesis of complete chemical process flowsheets. Mathematical design and heuristics from experience of past processes are the two main methods usually employed in process synthesis. Most approaches for new designs use heuristics based on studying reaction and separation systems in isolation. This thesis discusses the development of a new process synthesis systematic procedure and software that integrates a knowledge based system with Aspen HYSYS process simulator, HYSYS optimizer, Aspen Icarus economic evaluator, and databases, utilising knowledge from existing industrial processes to obtain design rules. The proposed generic superstructure for the synthesis and optimization of reaction-separation-recycle systems has been validated. To account for the non-ideal behaviour of reactors, modular simulation is used and an example of the approach is illustrated for a fluidized bed reactor. Preliminary work in customizing Aspen HYSYS to simulate new unit operation has been illustrated. A Visual Basic for Application (VBA) programming code has been developed to link the integrated knowledge based system (IKBS) to Aspen HYSYS. The prototype IKBS has been applied for the selection of reactor-separator-recycle systems for ethylene oxide, ethylene glycol, acetic acid and cumene manufacturing processes as case studies. A wide range of chemical reactors and separators were considered during the selection process and then elimination occurs at different levels leading to the best alternatives being selected for simulation, optimization and economic evaluation in the second phase of the IKBS for future development. The suggested alternative reactor-separator-recycle systems by the IKBS include currently used processes in addition to novel and recommended reactors/separators in industrial research. The proposed integrated knowledge based approach to chemical process flowsheet synthesis is expected to yield a cost effective design methodology for the petrochemical industry.