Statistical Modeling for Process Parameters of a Single-Screw Extruder in Betagro Group

  • พิชชากร ตันตระวาณิชย์ Kasetsart university
  • ภาณุมาศ อรุณเดชาวัฒน์
  • วิศว์ ศรีพวาทกุล
Keywords: Extrusion cooking, Statistical modeling, Data cleansing


Although the technology of producing the fish feed of Betagro Co. Ltd. is high, there is an issue that each employee might set the initial condition of variables differently.  This makes the company lost time and raw materials.  This research framework focuses on the independent variables, which are Feed rate (f) (3.8-4.5 ton/hr), Steam precondition (sp) (0.29-0.45 ton/hr), Water precondition (wp) (0.64-0.87 ton/hr), Steam barrel (sb) (0-0.4 ton/hr), and Water barrel (wb) (0-0.29 ton/hr).  These variables are independent.  They affect the dependent variables, which are Bulk density (b) (418-468 kg/cm3), Die pressure (recip_die) (20-26 bar) and Load (l) (44-54% full motor).  This research aims to explain the relationship between the independent variables and dependent variable.  This research collects 318 data.  Nevertheless, the raw data needs the processing of data cleansing to random sampling 3 times that recheck wrong data.  Furthermore, the data were normalized, removed the outlier, transformed, and de-duplicated to make them useful.  Therefore, the data remaining for generating the model is 262 data.  This research used Minitab18 to reach the goal by using the second-order polynomial in responses surface methodology (RSM).  The result of RSM showed that bulk density was increased by increasing feed rate and water precondition.  Load was increased by high feed rate, and steam precondition, while Die pressure was risen by high steam precondition and water barrel.  Finally, the independent variables include Feed rate, Steam precondition, and Water precondition impact the final product.