Targeting Batch Outcome using Predictive Control

Overview:

The paper will present a new methodology for using predictive control to improve batch process outcomes. Predictive control refers to a situation in which data collected during the execution of earlier process stages are used to retarget controllable variables during later process stages. Improvement is measured either by increasing output yield or by shifting the distribution of product outputs towards higher bin values, thereby increasing the value of a given output quantity. The technology includes a graphical method of presenting the logical flow of causality and functionality along a process. The paper describes how this methodology has been used in order to improve production outcomes in a batchbased system. The first part of the paper will present the technology, its pros and cons relatively to the traditional S88 without predictive control, and how the adaptive system is integrated within control systems. The second part of the paper will present results from a specific case study in which this technology has been implemented.