Condition diagnosis of a screw extrusion process through energy consumption curve monitoring

Traditionally, energy monitoring of processes was motivated by environmental concerns, disconnected from the actual process performance and reliability. However, energy and performance are intertwined, meaning a degraded system condition can cause power consumption changes. This paper presents an approach to condition monitoring, fault detection and diagnosis based on the calibration of energy operational curves, the relationship between a system’s power consumption and its operational performance. It differs from approaches that solely rely on the energy profiles over time, and instead extracts information from the coefficients of the relationship obtained by measuring the power consumption of the system against a sweep across one of its inputs. The approach was applied to a screw polymer extrusion machine, on which temperature faults and contaminated material were introduced. The proposed approach was able to detect a barrel temperature fault of a $\mathbf{1 0 \%}$ deviation as well as a contamination level of $15 \%$, all simply by studying the energy operational curve shifts. The approach offers a simpler option of instrumentation to diagnose equipment operational faults.