Satesh Sah (M. Tech. Awarded on 29/07/2022)


Designation: Research Scholar
Degree: M. Tech.
Fellowship: Fellowship (GoI)
Enrollment No.: 2020MMM003, Date: 08 Apr 2022
Supervisor: Debdulal Das

Academic Qualification

B. Tech. (ME), Techno International Newtown, Kolkata, 2020

Research Area

Topic: Processing-Structure-Property Correlations of Dual-Phase Steels using Machine Learning Approaches

M. Tech. (Specilization: Iron and Steel Technology), Awarded: July 2022


Title: Processing-Structure-Property Correlations of Dual-Phase Steels using Machine Learning Approaches


Abstract: The demand for advanced high-strength steel with requisite formability is increasing rapidly, specifically in the automobile industry. Using high-strength steel reduces the car body's weight and thereby ensures better fuel efficiency without compromising the safety requirements. Better formability is a prerequisite for near-net-shape processing and reducing the production cost. Therefore, the application of dual-phase steels in the automatic sector has been growing faster in recent years than ever. Achieving high strength without compromising ductility is still a challenge to materials scientists since these are two contradictory properties of a material. Understanding the interrelationships amongst the chemistry, processing factors, microstructural features, and various mechanical properties is an a-priori requirement. However, these correlations are often very complex; hence, artificial intelligence like machine learning techniques are increasingly used in this research domain. The recent studies mainly focus on constructing a model to relate one property of the material that gives the best fit.

In the present study, mathematical relations amongst the chemical composition, processing factors, microstructural parameters, and different mechanical properties of dual-phase steels are developed using various machine learning tools. Multiple linear regression (MLR), multiple non-linear regression (MNLR) with and with correlation matrix, and artificial neural network (ANN) have been employed considering relevant data extracted from the literature published in the last four decades. The multi-objective optimization has been used to find the critical values of alloying elements and processing parameters for superior strength and ductility combination.

It has been shown that elements like C, Cr, N, Mo, Nb, Ti, Ni, Cu, and B show higher dependency than other alloying elements on martensite volume. The hardness of dual-phase steels is mainly influenced by C, Mn, Ni, TIAT, and t; while both yield and tensile strength are controlled by C, Mn, Si, Cr, Nb, Ti, Ni, Cu, TIAT, t, type of heat treatment and martensite volume. For ductility of dual-phase steels, C, Si, Cr, V, Ni, Cu, TIAT, Def, type of heat treatment, and martensite volume are the most influencing parameters. Sensitivity analysis shows the positive effect of alloying elements and process parameters on martensite volume and properties like hardness and strength. The ductility shows the inverse impact on alloying elements compared to strength. The effects of alloying elements are better understood by simulation studies, where non-linear variations are observed in the volume of martensite and properties, both individually and in combination. The developed ANN models have been employed to obtain an optimum composition and heat treatment parameters for better strength and ductility combination using multi-objective optimization by genetic algorithm. The alloying elements like Ni, Mo, Nb, and V maximize the strength-ductility combinations of dual-phase steels. The obtained results provide new insight into developing high-strength dual-phase steels with good ductility and toughness.