Abstract
Out of specification in the production of Refined Bleached Deodorized Palm Oil (RBDPO) will cause recycling in palm oil refining process as well as zero production during that period. Hence, the ability to predict the output quality variables so that appropriate adjustment can be made in the process beforehand is very useful in minimizing the production cost and time consumption. In this study, Multivariate Statistical Process Control (MSPC) is used to develop a prediction tool in order to predict the output quality variables before the process even began and eventually reduce the utility cost as compare to traditional method of using tacit knowledge which is not very practical. Knowing product quality in advance, adjustment can be made at the specific unit operation and also to identify in which standard the quality variables belongs to, be it Palm Oil Refiners Association Malaysia (PORAM) standard, China or Vietnam. The MSPC method used in this study is Partial Least Squares Regression (PLS) which capable in finding the relationships between the process and quality variables of the palm oil refining process assuming that the data are linear. In simulation, data used are obtained from Lahad Datu Edible Oil Sdn. Bhd. (LDEO) with percentage of fatty acid (FFA), moisture content, and iodine value as both the process and quality variable which undergone pre-screening and pre-processing of data. This study also considers a constant sampling time to ensure the randomness of the data as well as the residence time in determining the relationships of the quality variables corresponded to the actual process variables. Simulation results obtained from developed correlation coefficient are evaluated and compared using Mean Squared Error (MSE). MSE results shows that PLS method is capable in predicting the output quality variables.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call