Condition monitoring of filtration subsystems in hot-mix asphalt plants

In today’s industrial landscape, the evolution of condition monitoring systems stands at the forefront of research, driven by the objective of reducing operational costs and minimizing environmental footprint. This thesis presents the work conducted during an internship at Marini-Fayat, which aimed to develop a condition monitoring system for the filtration subsystem of a hot-mix batch asphalt plant. This thesis provides an overview of the asphalt composition, the asphalt production cycle, and the main components of an asphalt plant. To identify the most significant anomalies, FMEA analysis has been employed. Starting from the plant fluidic model, a diagnostic cycle and a physics-driven diagnostic procedure, based on the concept of fluidic resistance, have been developed. The primary outcome of this study is a diagnostic algorithm that has been developed by employing the RLS algorithm and RBF neural networks, and it is based on the estimated static gain. The algorithm has been translated into ST code using MATLAB automatic code generation tools, and implemented into Beckhoff PLC. It will be tested on the plant to validate the methodology and acquire the necessary data to build a health indicator. Finally, extension of the diagnostic procedure to multi-parameters identification has been explored through the study of the frequency response of the system.