»
Fouling
Heat exchangers operating in cleaner conditions typically require less frequent maintenance compared to those in high-fouling environments.
Methodology Data collection The prediction of heat exchanger fouling requires identification of key operational parameters from both process and mechanical design aspects. Data acquisition of process parameters includes fluid properties, temperature and pressure profiles from limited number of sensors available at site. The collected data is systematically organized in Excel, enabling comprehensive simulation of cooling water network operating conditions by ICON. This simulated process parameters will be delivered to HTRI for heat exchangers thermal analysis. Additional equipment data, including engineering drawings and datasheets, are acquired for HTRI modelling purposes.
HTRI automation input The in-house HTRI automation server operates through a streamlined three-component input system.
• HTRI model files- validated heat exchanger models
• PI Excel files – process parameters extracted from ICON
• JSON configuration files- structured mapping files for process parameters into HTRI
The system incorporates robust error handling protocols. When computational anomalies are detected, the program automatically suspends execution and generates error messages for
user resolution. Upon successful completion of calculations, the automation server generates comprehensive output data in Excel, which is archived within the server for future reference and analysis. Key operational parameters analysed include thermal duty, fluid composition, fluid temperature and pressure profiles, flowrate, etc.
HTRI automation output processing The processing of HTRI output data follows a rigorous methodology to ensure data quality and reliability, the data was randomly selected for validation by engineers. Initial data preparation involves data cleaning to resolve missing values and duplicates. Comprehensive quality assessment using box plots in Azure, enabling the identification of outliers and understanding data distribution patterns.
Fouling prediction model development and validation strategy The model development phase employs a structured approach to ensure optimal predictive capabilities. It is crucial to examine the relationships between fouling and operating variables through correlation analysis. Heat transfer coefficient( UA) exhibits linear relationships with fouling, thus Pearson correlation analysis is applied to identify influential operating variables as heat transfer coefficient or fouling predictors. Then, selection of regression models below is utilized for UA or fouling prediction.
• Linear Regression – baseline performance assessment
• Decision Tree – captures non-linear relationships
• Boosted Decision Tree – algorithms for enhanced prediction accuracy
Figure 5. In-house HTRI automation launch page.
www. heat-exchanger-world. com Heat Exchanger World July 2025
47