Clustering analysis for production system design of emulsifier for biodiesel B30 based on digital business ecosystem

Abstract

Biodiesel in Indonesia is synthesized from palm oil into Fatty Acid Methyl Ester (FAME). In Biodiesel B30, FAME and diesel fuels cannot be homogeneous because FAME is hygroscopic and has a higher density than diesel fuels. This problem can be solved by adding emulsifiers. However, the emulsifier production process is currently limited to the laboratory scale. So, it is necessary to simulate the scaling up emulsifier production system and also the classification of the emulsifier product in the quality control section for this process production. Its integration processes and product are also not following the quality standard. This study aims to identify the relation of emulsifier formulation’s attributes then classify the quality according to the emission of NOx gas. The methods used to determine analyzed system requirements are using BPMN 2.0. The rules are then compiled from the existing emulsifier production system using Association Rules Mining (ARM), and the K-means algorithm will cluster the result. These rules can be used as a reference in taking influential attributes for emulsifier formulation. K-Means algorithm models the rules from the ARM into clusters where data in one cluster has the same characteristics and different characteristics from other clusters. The dataset used is hypothetical data from the formulation and quality testing of the emulsifier. This study’s final results are ten attributes that were approved in the emulsifier formulation and 4 clusters of emulsifier product quality based on NOx gas emissions and separated water layers.