Probabilistic Thermal Modelling of AdBlue System Using Bayesian Networks

In Diesel engine exhaust after treatment system (ATS), Nitrogen Oxides (NOx) emissions control is achieved via Selective Catalytic Reduction (SCR) in which AdBlue or Diesel Exhaust Fluid (DEF) plays vital role. But AdBlue freezes below -11°C due to which in cold climate conditions system performance becomes critical as it affects efficiency as well as overall performance leading to safety and compliance with emission standards issue. So, it is essential to have a probabilistic thermal model which can predict the AdBlue temperature as per ambient temperature conditions. The present paper focuses on developing Bayesian Network (BN) based algorithm for AdBlue system by modelling probability of key factors influencing on its performance including AdBlue temperature, Ambient temperature, Coolant temperature, Coolant flow, Vehicle operating conditions etc. The BN Model predicts and ensures continuous learning and improvement of the system, based on operational data. Methodology proposed in the paper aims to demonstrate a probabilistic model that captures the interactions affecting the AdBlue system's thermal behavior.