Determination of the most significant manufacturing process parameters can be very valuable in improving product quality and plant reliability. Given the huge amount of recorded data and complexity of chemical plants, data mining is crucial to uncover information, knowledge, and patterns from the historical data.
This presentation describes how to facilitate decision making process in a polypropylene plant by using data mining techniques such as variables screening and modeling. High liquid hydrocarbon level within the polymer product is a chronical safety issue in polypropylene plants. Since the high hydrocarbon level can often be the cause of either plant shutdowns or reducing the production rate, understanding the root cause of this problem has a significant impact of the plant reliability and economics.
Data visualization and multivariate analysis techniques are used to identify relative significances of process variables and evaluate their effects on the hydrocarbon level. As a result of this study, new operational trials are proposed and tested in the plant. The result of trials is used to study the impact of changing in significant variables on reducing hydrocarbon level.
See more of this Group/Topical: Topical A: 2nd Big Data Analytics