Research Update: Predicting Heat Transfer Coefficients with an Artificial Neural Network

Artificial neural networks (ANNs) allow the solution of hard problems such as 

• determining important parameters affecting crude oil fouling
• analyzing operating data to troubleshoot heat exchangers
• developing correlations through multivariate nonlinear regression of complex data

To evaluate a neural network model for heat transfer coefficients, we used a multilayer, feed-forward ANN with shellside vacuum condensation data from the Low Pressure Condensation Unit (LPCU) and the Multipurpose Condensation Unit (MCU). In the webinar, we describe this type of neural network and discuss the findings of this study. We also highlight challenges and identify possible future uses of neural networks at HTRI.