Data-Driven Bayesian Modeling for Heat Exchanger Monitoring

• Grant number: 23/13794-7
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: November 01, 2023
• End date: November 30, 2025
• Field of knowledge: Engineering - Mechanical Engineering - Mechanics of Solids
• Principal Investigator: Samuel da Silva
• Grantee: Vitória Batista Godoy
• Host Institution: Faculdade de Engenharia (FEIS). Universidade Estadual Paulista (UNESP). Campus de Ilha Solteira. Ilha Solteira , SP, Brazil
• Associated scholarship(s): 25/07225-5 - Data-Driven Bayesian Modeling Using Approximate Bayesian Computation for Heat Exchanger Monitoring, BE.EP.IC 24/05412-0 - Data-Driven Bayesian Modeling Using Approximate Bayesian Computation (ABC) for Heat Exchanger Monitoring, BE.EP.IC

Abstract

Heat exchangers are essential pieces of equipment for many industrial processes and, in general, must be monitored in the field with sensors in order to diagnose their operating status with safety and energy efficiency. However, it is vital to build numerical replicas from this data that communicate with actions control in the physical plant, allowing simulations in a virtual environment, known as digital twins, providing more detail and information for diagnostics and decision-making. In this sense, data-driven Bayesian calibration tools have been widely used. The aim of this research is to get the student in this area by proposing a digital twin for a shell-and-tube heat exchanger, in order to monitor and diagnose its condition. The fellow will implement the parameter calibration of a simple heat exchanger model using the log mean temperature difference (LMTD) method, but assuming that the global heat transfer coefficients are random variables with known a priori distributions. The Markov Chain Monte Carlo (MCMC) method will be used to sample the distributions of the parameters to be calibrated in the tests to be carried out. In addition, a study of the global sensitivity of the parameters will be carried out using Sobol indices. The result of the project will be the verification of an inverse reduced-order model to obtain information on the state and operation of the heat exchanger, based entirely on the historical measurement data of the temperature and flow variables.