In industries manufacturing products such as the pharmaceuticals, dyes, polymers, bio-technical products and food products, batch processes are very important. In these processes improving production operations involves the design of control mechanisms and components at different levels. Almost invariably regulatory control is involved at the low level. At the next level, supervisory control is important. Problems at this level range from the optimisation of production schedules to the automation of production recipes and the detection and an automated response to complex alarms. The former class of problems has been relatively well studied, particularly for large scale continuous manufacturing. Interest in the latter category is relatively recent. One approach has been to call upon results from artificial intelligence research. Whilst much has been achieved and several general purpose software packages such as G2 exist, development is hampered by the lack of a simple and robust model for describing the dynamics of batch processes subject to both regulatory and supervisory control. The theory for describing a batch process at an event driven level is now beginning to emerge from the literature on discrete event dynamic systems. However, much of this research has been directed at problems for which there is no corresponding underlying regulatory control level. Hybrid models which capture both the discrete event dynamics as well as the continuous regulatory control underneath are just beginning to emerge. A good indicator of this is that the most recent World Congress of the International Federation on Automatic Control, in July 1996, had several sessions devoted to this subject whilst the previous Congress, held 3 years earlier, had very few papers on this topic. At NUS, we are seeking to initiate research and development work in this area. To systematically integrate sequential and alarm control systems with low level feedback control loops, conventional dynamic models utilizing differential or discrete equations are not adequate. The system contains event driven dynamics : the state changes only at discrete instants of time in response to particular events. These events may result from equipment malfunctions, unexpected process reactions or simply rapid changes in market conditions. Current supervisory control design techniques are largely ad hoc. Developing a supervisory control system is a painstaking exercise building on collective experience for specific processes. The consequence is that neither completeness nor consistency of the supervisory control system can be guaranteed. The supervisory control response will depend very much on the particular programmer's understanding of the process. We seek to develop a methodology for building hybrid models appropriate to the process industries which can alleviate some of these difficulties. Many modeling paradigms have already emerged but none have been systematically tested and evaluated on a plant. A reliable hybrid model would be valuable for the systematic development of integrated supervisory control procedures. To enhance the usefulness of this project, we would like to invite industrial partners to participate in this research. We would particularly welcome companies with an interest in batch process control or supervisory control to contact us for discussions on directions for work in this area in Singapore.