Real time batch process optimization within the environment of the flexible recipe

Abstract

Abstract Batch processes are normally thought to operate using a traditional fixed recipe. However, the fixed recipe has to be approximately adapted in a rather unsystematic way depending on experience and intuition of operators. Therefore, the concept of flexible recipe seems to be the adequate way to rationalize and systematize the adjustment procedure. In the flexible recipe context, the term recipe is used in a more abstract way by referring to a selected set of adjustable recipe items that control the process output. In the present work, a framework for batch process real time optimisation considering the flexible recipe concept is presented. As soon as a deviation is detected, the control recipe can be readjusted. Deviations are only detected at the process state assessment. The process state assessment compares the operating conditions at the sample time with the expected operating conditions that will depend on the initial conditions. For this reason and in order to perform this comparison, a predictive model is necessary. Then, a corrective model is required to adjust the control recipe. This second model describes the ultimate effect of the values measured at the time of the process state assessment as well as any run-time corrections applied during the remainder of the processing time.