New Framework for studying High Temperature Tribology (HTT) Using a Coupling Between Experimental Design and Machine Learning

Authors

• Mohamed Kchaou Department of Mechanical Engineering, College of Engineering, University of Bisha, P.O. Box 001, Bisha

DOI: https://doi.org/10.30678/fjt.142331

Keywords:

hot temperature, tribometer, wear, mechanisms, characteristics, artificial intelligence, new approach, coupling

Abstract

High temperature tribology (HTT) is considered to start at a minimum temperature of 300°C, where organic base oils and polymers begin to decompose, up to a temperature of 1200°C. In this area of application, a tribological test is typically performed under dry or solid friction, unless a solid lubricant is used, as most lubricants oxidize or decompose when exposed to these extreme temperatures. This is the case of hot forming tribology. Therefore, specific tribometers have been developed to study the tribological behavior, wear modes, friction-wear mechanisms, and other tribological aspects in the case of contact workpiece-forming tools. However, as the interdisciplinary character of tribology represents great opportunities, but also a huge challenge to well study the tribological behaviour of materials and systems, particularly at hot temperature, results are well impacted by the proposed hypothesis, the limited number of parameters to study, the dynamic behaviors of solicitations, etc. A new approach to overcoming its limitations remains a necessity in modern research. After giving general definitions of the notions of tribological system and tribometer, an assessment of the different configurations of these test benches is established. The presentation of the tribometers is organized according to the configuration adopted. The study is based on the identification of the “originality” of the benches and the limitations of the approaches used in the study of hot temperature sliding contact. The framework of the coupling between experimentation and machine learning is presented. Different scenarios are discussed in order to develop new approaches/methods of collaboration between the design of experiment, numerical development, and ML algorithms.