Time dependent Pinch analysis with sensor data and unsupervised learning algorithms

Abstract Heat integration is essential for energy efficiency and operational cost reduction in energy-related applications. Pinch Analysis a heat model that has attracted lot of attention over the past few decades. Although method been successfully applied to many industrial processes, it also shown have limitations. One these limitations only considers average values physical quantities process, therefore not suitable studying dynamic processes undergo variations their operating conditions. Attempts made improve circumvent this limitation, but they found oversimplify result unrealistic solutions. The purpose work use sensor data unsupervised learning algorithms extend processes. Specifically, we time series segmentation detect changes process conditions clustering group similar segments. approach real case. We compare performance exchanger network with without analysis. solutions obtained our methodology improves waste recovery by 20% while reducing 31%.