Supriyo Ghosh - Research

Tusas: A modern computational framework for microstructure evolution toward exascale

The Phase-field method is one of the most powerful techniques to simulate microstructure formation in metal alloys. Software development for phase-ﬁeld simulation is an emerging need for multi-physics materials modeling capability. Also, the software will be highly impactful in this high-throughput, data-driven era with the ultimate goal of reproducible research and code reuse. We developed a fully-coupled, fully-implicit approach for phase-field modeling of solidification dynamics and microstructure formation in metals and alloys. Our simulation approach consisted of a finite element spatial discretization of the fully-coupled nonlinear system of partial differential equations treated implicitly by the Jacobian-free Newton-Krylov method with physics-based preconditioning. This approach allowed us to use timesteps larger than those restricted by the traditional explicit Courant-Friedrichs-Lewy maximum timestep limit. Also, our approach was algorithmically scalable due to an effective preconditioning strategy based on algebraic multigrid and block factorization. We implemented this approach by introducing a new open-source phase-field simulation framework, Tusas, developed at Los Alamos National Laboratory. We analyzed the numerical performance of Tusas in terms of algorithmic scalability while demonstrating the computational performance in terms of parallel scalability on emerging heterogeneous supercomputing architectures (right figure). We have demonstrated ideal strong and weak scaling on up to a billion unknowns and thousands of GPUs on the two fastest supercomputers (summit and sierra, right figure) in the USA. In addition, we demonstrated the efficacy of Tusas using benchmark phase-field simulations of dendritic solidification in metal alloys under additive manufacturing conditions (left figures, 2D and 3D dendrites). The benchmark simulations were validated extensively against related experimental measurements and physics-based analytical model predictions (middle figure). Currently, Tusas plays a critical role in studying solidification dynamics and microstructure formation during metal casting and additive manufacturing. Tusas is already available through a GitHub repository https://github.com/chrisknewman/tusas. Refer to publication article #21 for more details.

Machine Learning (ML) and Uncertainty Quantification (UQ) for 3D Printing

Machine learning approaches have the potential to predict the essential process and microstructure signatures rapidly at an acceptable level of accuracy. The uncertainty in process parameters in the macroscale leads to variability in microstructural features. Once the process and microstructure information are available from experiments and simulations, the uncertainty sources are analyzed in a forward approach followed by an iterative inverse approach to update the critical input parameters to reduce the uncertainty toward a proactive control of the macroscale quantity of interest (QoI). The left figure in below illustrates the overall Uncertainty Quantification (UQ) framework to quantify the 3D printing-microstructure-properties-performance correlation. The potential of a Gaussian Process (GP) surrogate model is shown (on the right) toward successful prediction of the phase-field simulated composition distribution of the primary segregation element that results during solidification of an Inconel superalloy melt-pool. For more, refer to publication articles #16 and #19.

Rapid, Non-equilibrium Solidification in Metal Additive Manufacturing

During an additive manufacturing process, parts are built layer-by-layer by rastering the feedstock (wire/powder) and laser/electron beam across the substrate material that is followed by complex melting, solidification and solid-solid processes. Solidification occurs rapidly in the trailing edge of the molten pool, resulting in columnar and/or equiaxed microstructures which determine the properties of the deposited microstructures. A typical temperature distribution during additive manufacturing is shown below. Referring to this temperature profile, γ (Ni alloys) or β (Ti alloys) cells/dendrites solidify directionally and grow perpendicular to the solid-liquid boundary approximated by the solidification isotherm with a fixed temperature gradient and at a interface velocity. For more, refer to publication articles # 5, 7, 9, 10, 11, 12, and 15.

Phase Separation in Complex Mixtures

During spinodal decomposition, the liquid spontaneously phase separates to its constituent solid phases (bottom left figure). Phase separation in complex mixtures, which may include alloys, polymers, and metallic glasses, is modified by a pre-existing "foreign phase" (mobile or immobile) in the matrix in the form of spherical particles or any arbitrary shape with sizes ranging from tens of nanometers to microns. Phase-field simulations are performed to study phase separation of a homogeneous liquid in the presence of immobile spherical particles with "selective wetting" for one of the components, leading to target, continuous, lamellar, or droplet patterns, depending on interface energy, volume fraction, and interparticle distance of the bulk phases. These patterns find applications in designing nanocomposites and polymer blend films. For more, refer to publication articles #6 and #20.

Eutectic Solidification: Composite Microstructure Patterns

During eutectic solidification, a liquid of eutectic composition (taken from phase diagram) solidifies into composite structures of two solid phases which grow by a mutual solute diffusion in between the liquid regions in front of each phase, resulting in either lamellar (shown below) or hybrid morphologies, depending on volume fraction of the component phases. The lamellar microstructures shown below were simulated using a multi-phase-field model to study the influence of surface energy anisotropy towards the orientation of the solid phases in bulk composites. In the absence of anisotropy, a complex network of lamellae or a labyrinth (left figure) forms. A weak anisotropy in solid-solid boundaries leads to regular lamellar arrays with/without defects along a preferred orientation, which is a minimum energy direction. The simulations agree with experimental observations. For more, refer to publication articles # 1, 2, 4, and 8.

Grain Boundary Orientation/Properties

When liquid solidifies into two grains of the same phase that grow next to each other, a grain boundary with a groove form in between them in response to the mechanical equilibrium of grain boundary and solid-melt boundary tensions. Below are some results from multi-phase-field simulations that reproduce some experimentally observed orientations such as angular or vertical. Solidification rate was varied form zero to morphological instability threshold (when a planar solid-liquid interface becomes cellular/dendritic shape) and beyond to explore the possible orientations of grain boundaries, which varied as a function of surface tension, mobility, and growth velocity. These informations are useful to calculate grain boundary energetics and kinetics. For more, refer to publication articles # 3 and 17.

Tusas: A modern computational framework for microstructure evolution toward exascale

Machine Learning (ML) and Uncertainty Quantification (UQ) for 3D Printing

Rapid, Non-equilibrium Solidification in Metal Additive Manufacturing

Phase Separation in Complex Mixtures

Eutectic Solidification: Composite Microstructure Patterns

Grain Boundary Orientation/Properties