A dynamic approach to energy efficiency estimation in the large-scale chemical plant. (1st March 2019)

• Record Type: Journal Article
• Authors: Zhu, Li; Chen, Junghui
• Is Part Of: Journal of cleaner production. Volume 212(2019)
• Page Start: 1072
• Page End: 1085
• Publication Date: 2019-03-01
• Subjects: Energy efficiency estimation models -- Industrial energy efficiency -- Nonlinear modeling method; Factory and trade waste -- Management -- Periodicals; Manufactures -- Environmental aspects -- Periodicals; Déchets industriels -- Gestion -- Périodiques; Usines -- Aspect de l'environnement -- Périodiques; 628.5
• DOI: 10.1016/j.jclepro.2018.11.186
• Languages: English
• ISSNs: 0959-6526

Abstract

Abstract: With the increasing pressures from the energy price and environmental protection, large-scale chemical plants pay more attention to the implementation of energy efficiency estimation to improve its economic benefit and environmental performance. Because of the stochastic and dynamic characteristics of the actual data, traditional estimation methods fail to satisfy the requirement of real-time evaluation. To cope with this limitation, a novele nergye fficiency estimation method combiningj ust-i n-t ime (JIT) learning ands ubspacem odeli dentification (SMI) with noise elimination, called e-JITSMI method, is proposed. First, the state space model is constructed to describe the dynamic performance of production processes. This integration method can select the appropriate sampling data, estimate noise effect, and build the corresponding dynamic model. With the built model, not only are the relationships between production and supplied energy built, but the energy efficiency tendency is also predicted at the next moment. In addition, with the arrival of the new sampling data, the dynamic evaluation model is automatically updated. The effectiveness and accuracy of the proposed method are demonstrated through a practical large-scale chemical process. The results present the average accuracy of energy efficiency prediction can reach 88.9% and the tendency of energy efficiency is 100% correct even if the working conditions change. Highlights: E-JITSMI is proposed to estimate energy efficiency and build just-in-time learning. e-JITSMI can evaluate the energy efficiency for large-scale plants in real time. E-JITSMI yields consistent energy efficiency in a noise effect situation. Energy efficiency tendency is predicted by e-JITSMI at the next moment. Practical ethylene production validates the effectiveness and accuracy of e-JITSMI.