MINI DISTILLATION COLUMN CONTROL USING MODEL PREDICTIVE CONTROL

• Main Author: SAPTO HANDOKO (NIM : 23215004), ANTON
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/21168
• Institution: Institut Teknologi Bandung

Distillation column is a very important equipment in the chemical industry, which requires large thermal energy for 30% until 40%,. So, it requires optimal control to obtain an efficient production process. The model predictive controller is an optimal controller which can solve the operational problems. In this research, control of the ethanol water separation process in the mini distillation column of batch type is performed in Electrical Engineering Honeywell Laboratory in ITB. The dynamic of ethanol concentration on the reflux valve is modeled as a linear system with time delay at certain operating point. Model predictive control with 5 horizons are implemented in Arduino Mega 2560 with 1 minute sampling time. From several experiments, with 1400 W heating power, and 30 minutes operational time, the result of ethanol concentration is obtained with different tracking error, at least 2%. A higher tracking error can be found due to various things, such as untested ethanol concentration sensors. To improve model prediction, the implementation with 3 minutes of sampling time and 5 horizons are also performed with unsatisfactory results due to the sensor's inaccuracies. In general, the concentration of ethanol can be regulated by a model predictive controller. It shows that the linear model used is quite representative. Improved performance of the control system requires an accurate and calibrated concentration sensor.