Coolant Characterization and Predictive Monitoring System for Heavy-Duty Trucks using Machine Learning

Ivan Khai, Teo Yen and Pei Yi, Lee and Leong Seng, Wong and Clarence, Chin and William, Liang and Siong Fong, Sim (2025) Coolant Characterization and Predictive Monitoring System for Heavy-Duty Trucks using Machine Learning. Malaysian Journal of Chemistry, 27 (6). pp. 13-28. ISSN 2550-1658

PDF Coolant Characterization.pdf Download (1MB)

Abstract

Over 50% of engine failures are attributed to poor coolant system maintenance. This study reports on a fleet test of automobile coolants and the development of a monitoring system to evaluate coolant performance. The fleet test was conducted on six trucks, where parameters such as pH, reserve alkalinity (RA), ethylene glycol (EG) content, and functional characteristics of the coolant were measured over eight months, and then subjected to regression learning. A coolant monitoring system was also developed to record and visualize coolant performance. The tensile strength of the radiator hose was examined to assess the impact of coolant on hose integrity. The study identifies pH as the most suitable indicator of coolant health against mileage with its stable degradation trend and clear threshold, though insensitive to coolant dilution. The C-O vibration of EG, as detected by Fourier Transform Infrared (FTIR) analysis, corresponded well with EG content determined using a refractometer. There was no evidence that coolant affects the tensile strength of the radiator hose. The Gaussian Process Regression (GPR) model was used to predict coolant degradation and estimates the optimal replacement mileage at of 111,473.65 km. This study introduces a data-driven coolant management approach to enhance reliability, cut maintenance costs, and reduce downtime.

Actions (For repository members only: login required)

View Item