ABSTRACT
Batch processes are well suited for the manufacture of low-volume high-value products and are often used for the flexible manufacturing of multiple related products in the same facility. Batch processes are usually preferred for production volumes below 10,000 Mt/year, whereas continuous processes are predominantly used when the production volume is in the order of 100,000 Mt/year. Batch processes are
“73605_C015” — #2
the production scheme of choice for the pharmaceutical, biotechnology, specialty chemical, consumer products, and microelectronics industries. The production of these high-value-added chemicals contributes a significant and growing portion of the revenue and earnings of the chemical process industries. Batch processes are heavily used in the pharmaceutical industry where isolation and product consistency are required for reasons of safety and sterility. Additionally, the shutdown and startup of high-volume production continuous processes under economic and safety constraints can be treated as a batch control problem. Despite their widespread application, batch recipes and control strategies are often designed on an empirical basis in industrial practice due to costs for developing process models and the limited quantity of online instrumentation available. Over the last decade, increased availability of suitable models and measurement devices has lead to an increased interest in the dynamic (online) optimization, which has the potential of yielding significant performance and quality improvements. A series of potential issues must be addressed for a successful application of optimal batch control strategies. Usually the initial conditions are only roughly known, most states are unmeasured, and disturbances and model uncertainties are present. These problems must be taken into account during control systems design, because even small disturbances and model uncertainties can lead to significant performance degradation in batch processes.
