Preprocessing batch profiles for statistical process control

Batch processes can be monitored and analyzed by applying SPC techniques. The primary motivation behind their use is to identify and study process variation in an effort to improve process efficiency and the quality and consistency of a product. Conventional SPC techniques are quite useful in identifying process variation, but they alone provide little indication as its type or source. In this study, a technique is developed for analyzing batch profiles. It is based on the same principles as a convention SPC analysis, but it is capable of providing more quantitative as well as qualitative information. It uses a preprocessing step to analyze batch profiles before applying any SPC techniques. The preprocessing step is made up of a least square analysis that provides deterministic clues as to the source of variation. This analysis is capable of quantifying the effects of variation on a profile's initial condition, magnitude, and speed of development. It produces an adjusted profile that can be used to increase the sensitivity of an SPC analysis, and also generates a set of parameters that can be used in correlation studies and easily stored in a historical database.