Development of energy efficiency principal component analysis model for factor extraction and efficiency evaluation in large-scale chemical processes

In a large-scale chemical plant, it is important to evaluate the energy efficiency (EE) of production to improve the production process and make production decisions. Essentially, finding the relationship model is the foundation of EE evaluation. Given the requirements of universality and practicability, the data-driven model is widely used to describe the variable relationships. However, the variables stored in the data bank may be redundant, and some variables contain disturbances in the large-scale chemical process, increasing the complexity of the model establishment. In this paper, a new EE factor extraction and EE evaluation method based on principal component analysis (PCA) (EEPCA) is proposed to enhance the accuracy of the EE values. By three stages (noise term estimation, model establishment, and model variable selection) in EEPCA, the accurate relationship models of utilized energy mediums and chemical products are established. On the basis of the built models, the EE of the chemical processes is evaluated and inferred. The effectiveness and practicality of the proposed method are demonstrated via a simulated process and a practical ethylene production.