Development of an Extraction Column using Machine Learning for the Prediction of Feed and Solvent to Obtain the Desired Extraction

Abstract

The phrase "Industry 4.0" describes the fourth industrial revolution, which is characterized by the usage of digital technology to improve automation, connectivity, and efficiency across various industrial processes. The optimal column designs, suitable solvent selection, and extraction yield prediction are all aided by artificial intelligence (AI). Integration of AI technology with extraction columns is a significant step forward for industrial processes, bringing in a new era of intelligent systems that drive previously unheard-of increases in quality, efficiency, and environmental responsibility. The choice of acetone and water as the input solvents and Methyl Isobutyl Ketone (MIBK) as the extracting solvent results in an ideal configuration for the machine-learning design of an extraction column. With acetone's enhanced solubility and miscibility properties and water's versatility as a solvent, the two work well together. The extraction column was created using Python and concepts and methods based on chemical properties as well as machine learning. The intended extraction column may be implemented with the help of the Python code that is supplied. Based on parameters entered by the user, including extract temperature, raffinate temperature, extract mass flow rate, and extract temperature, it forecasts the temperatures and mass flow rates of acetone and water. By considering the properties of mass flow rates and temperatures, the projections ensure physical feasibility. Utilizing controlled data for training, the model applies a linear regression approach. The findings include the mass flow rate of acetone, the temperature of the water, and the water mass flow rate. The integration of state-of-the-art machine-learning techniques and Industry 4.0 is made possible in this work. This holds the promise of increased process optimization and efficiency in chemical extraction processes.