About us
Research
Deformation behaviour of duplex stainless steels Duplex stainless steel are a highly alloyed two-phase material with nearly equal austenite (fcc) and ferrite (bcc) phase fractions. They are extensively used in oil and gas industries. In collaboration with Sandvik, we are studying the deformation behaviour of SAF906. We aim to understand (i) deformation mechanisms and texture evolution of individual phases (ii) strain partitioning among the phases (iii) effect of orientation relationships on the texture evolutions and (iv) the mechanical anisotropy evolution.
Deformation twinning in hcp materials
Deformation twinning is a strain accommodating mechanism along with crystallographic slip in many crystal systems. They are more commonly encountered in crystals with lower symmetry and with access to less than five independent slip systems such as hexagonal closed packed and tetragonal crystal systems. At a fundamental level, we are trying to address the role of local grain interactions on the nucleation and growth of deformation. We are also studying the interaction of deformation twins with precipitates in precipitation strengthened Mg alloys and their role on the tension/compression asymmetry. We are also looking into the long standing unanswered question about the role of grain size on the twinning behavior.
Virtual testing of materials: multiscale linking Finite element simulations of metal forming operations require constitutive models of underlying metallic sheet materials capable of realistically capturing the plastic anisotropy due to the underlying crystallographic texture. For the sake of computational efficiency, the material response is implemented in the form of yield criteria. These yield criteria often have a large number of material parameters that need to be obtained from a series of complex mechanical tests. Often such tests are either inaccessible or not easy to perform. Virtual material testing can then be used as a framework where one can perform a series of numerical experiments on statistically representative microstructures with high fidelity material models such as crystal plasticity models to obtain the required parameters for the anisotropic yield criteria. We write user material subroutine VUMAT to be implemented in commercial finite element software, ABAQUS for anisotropic yield criteria. The material parameters for the yield criteria are obtained using crystal plasticity model. Going forward, we plan to develop a more comprehensive platform where users can choose the right material model at the microstructural scale and the appropriate yield criteria at the sample scale.
Machine learning assisted prediction of mechanical response of materials. Recently, we have started working on applying data science tools to solve problems related to materials. In particular, we have been looking into the quantitative prediction of processing-structure-property correlations. We are currently tackling the issue of the dimensionality curse of big data encountered in materials science with particular emphasis on the intrinsic dimensionality of microstructure data. We are also looking at ways to apply deep learning algorithms to build reduced order models to accurately predict the local stress/strain distributions in two phase materials. We hope to develop a data enabled framework to predict the initiation and damage in engineering materials accurately.
Grain boundaries Grain boundaries are essential features of microstructures. A better understanding of the mechanical response of GBs is required if we are interested in understanding mechanics of microstructures. We have been looking at two aspects

What is the intrinsic stress state of a grain boundary ?
What is the width of the grain boundary influence zone?
Funding
image image image image image image