Application of Machine Learning in Failure Prediction of Brazed-Aluminum Based Heat Exchangers

Brazed aluminum plate-fin heat exchangers (BAHX), extensively used by the natural gas liquid recovery and gas processing industry, are subject to a uniquely stressful and harsh environment that involves large thermal stresses due to steady-state and transient thermal loads, mechanical loads, contaminants, corrosion and exposure to low temperatures. Despite careful design for unlimited life, several pre-mature failures in different natural gas processing plants have been happened which has resulted in significant loss of revenue and serious life threats. In this work, we used finite element analysis to explain potential reasons for failure occurred at two failed BAHX. Since this kind of modeling has limited utility and a stronger tool for prediction of failure has to be developed, we train an artificial neural network which can predict failure by distinguishing between healthy operating conditions and unhealthy operating conditions. An excellent accuracy was obtained for the current available dataset. However, since available dataset was extremely limited, we believe training should be performed on a much larger dataset to have reliable results.

Name thesis:degree_name

Master of Science

Level thesis:degree_level

Masters

Discipline thesis:degree_discipline

Mechanical Engineering

Grantor

University of Houston

Year dc:date.issued

2019

Author dc:creator

• Rahmati, Amir Hossein

Advisor dc:contributor.advisor

• Sharma, Pradeep

Committee members dc:contributor.committeemember

• Kulkarni, Yashashree
• Nakshatrala, Kalyana Babu

• Brazed aluminum plate-fin heat exchangers
• Machine learning

Statement dc:rights

• The author of this work is the copyright owner. UH Libraries and the Texas Digital Library have their permission to store and provide access to this work. Further transmission, reproduction, or presentation of this work is prohibited except with permission of the author(s).

Language dc:language.iso

eng

Handle dc:identifier.uri

https://hdl.handle.net/10657/5731

OAI identifier oai:identifier

oai:uh-ir.tdl.org:10657/5731