Machine Learning Embedded Reduced Order Modelling (NPOD) of Complex Droplet Dynamics

Modeling complex droplet dynamics—such as deformation, breakup, and coalescence—is computationally demanding using full-scale simulations. Reduced Order Modeling (ROM) simplifies these systems, but traditional methods often fail to capture nonlinear behaviors accurately. Integrating machine learning (ML) with ROM enhances predictive capability by learning complex temporal and spatial patterns from high-fidelity data. Techniques like LSTM networks and physics-informed neural networks (PINNs) enable accurate, efficient simulations while preserving physical laws. This hybrid ML-ROM approach offers significant reductions in computational cost, time, and CPU memory and is well-suited for real-time predictions, control, and optimization in multiphase flow applications. 

High-fidelity experiments (FLI, Shadowgraph, PLIF, VLIF, ILIDS) and theoretical model development on sprays and atomization in rotary atomizers

Rotary atomizers are used in small-scale gas turbine engines, chemical processing, pharmaceuticals, spray coating, desalination, and many more applications based on design, speed, and liquid feed range. So, we are trying to explore the underlying physics of different breakup modes of liquid, atomization regimes, droplet sizes in sprays, and many more using different high-fidelity experiments to achieve better spray quality and next-gen atomizer design.  

Droplet dynamics upon impingement on different surfaces: Numerical, theoretical, and thermodynamical model development

The impact of a single droplet on various surfaces is ubiquitous while its omnipresence could be observed in almost all industries ranging from pharmaceuticals, chemical, spray drying and coating, ink-jet printing, jet cleaning, self-cleaning surface to anti-icing, forensic science, and in many more. Presently, we are trying to develop numerical, theoretical, and thermodynamical models the droplet impingement on various surfaces in a quasi-steady medium. The shape of the droplet varies from spherical to ellipsoid (by varying the ratio of major to minor axis) to investigate the effect of droplet shape on its spreading and retraction dynamics. 

Spray Combustion in Turbo-Swirl Atomizers

Spray combustion in turbo-swirl atomisers involves the efficient mixing and burning of liquid fuel injected into a high-speed, swirling airstream. The atomiser imparts strong tangential velocity to the air, creating a recirculation zone that enhances atomisation and fuel-air mixing. This results in fine droplets, improved evaporation, and stable combustion. The swirling motion also promotes flame anchoring and reduces emissions by enabling more complete combustion. Turbo-swirl atomisers are widely used in gas turbines and aerospace engines due to their compact design, high combustion efficiency, and ability to handle varying operating conditions while maintaining low pollutant formation and reliable flame stability.

Solid Propellant Combustion for Rocket Propulsion

Solid propellant combustion is a key process in rocket propulsion, where a solid fuel mixture combusts to produce high-pressure, high-velocity exhaust gases that propel the rocket forward. The propellant typically consists of a fuel, oxidizer, and binder, which are chemically formulated to provide optimal energy release. During combustion, the propellant undergoes a complex chemical reaction, releasing heat and gas, while the burn rate is influenced by factors like pressure, temperature, and surface area. This controlled combustion process ensures a stable and predictable thrust, crucial for mission success. Solid propellant rockets are commonly used in military, space exploration, and satellite launching applications.

Reduced Order Modeling of Marging flames

Reduced Order Modeling (ROM) of merging flames aims to simplify the complex, high-dimensional behavior of interacting flame fronts into a lower-dimensional framework that captures essential dynamics. Researchers can extract dominant flow and combustion features from detailed simulations or experimental data by applying techniques such as Proper Orthogonal Decomposition (POD) or Dynamic Mode Decomposition (DMD). This enables efficient prediction and analysis of flame merging phenomena, including heat release, instabilities, and flame topology changes, at a fraction of the computational cost. ROMs are especially valuable for real-time control, design optimization, and understanding of flame behavior in turbulent or confined combustion systems.

Droplet Combustion

Droplet combustion is the process where a liquid fuel droplet burns in a surrounding oxidizing environment, typically occurring in stages: heating, evaporation, mixing, and combustion. As the droplet is heated, it vaporizes, forming a flammable mixture with the surrounding gas. This leads to the formation of a flame envelope around the droplet, where combustion occurs. The process is influenced by factors such as ambient temperature, pressure, fuel volatility, and droplet size. Droplet combustion is fundamental to understanding spray combustion in engines and gas turbines, where numerous droplets burn simultaneously. Studying it helps improve fuel efficiency and reduce pollutant emissions.