The Indian Institute of Technology (IIT), Jodhpur has collaborated with Algo8 AI Private Limited to design a data pipeline build-up and machine learning model for heat exchangers of oil and petroleum refineries. The research team of Jodhpur including the head and assistant professor of the department of chemical engineering, Pradip Kumar Tewari and Angan Sengupta with a team of postgraduate students are providing consultation services for this project.
A common event in refineries is the fouling of heat exchangers. These units need periodic maintenance via a complete shutdown or by bypassing the unit in the network. For the efficient functioning of these heat exchangers, controlled and schedule-based maintenance is required. This problem needs a foundational understanding of the applications of process engineering and data modelling techniques for a wide
The research team has designed a database model for the control and scheduled maintenance of heat exchanger networks in the refineries. The solutions provided include a comprehensive and verified predictive cleaning schedule for pre-heat trains, a graphical interface to different refinery units for control mechanism and cleaning and purging data quality matched with industry specifications.
Speaking about the project, Sengupta said, “This particular project depicts that the present-day chemical engineering, of which computation for establishing industry 4.0 is an integral part, has traversed a long way from traditional chemical engineering and is capable to meet the new challenges of various process industries. To this end, the department of Chemical Engineering at IIT Jodhpur is well equipped.”
The model cannot only trace the problems faced in regard to heat exchangers but can also be applied in any petroleum and other allied industries. A few other main aims of the project include condition monitoring of heat exchanger trains, assessing the network impact of each heat exchanger unit, a graphical user interface to provide to visualize the predicted results and to verify the accuracy of the model, model-based improved asset availability with in-time warnings and Improved shutdown turn-around time among others.