2021 - Current (During PhD)

• Designed a five loop oscillating heat pipe with a true adiabatic section, instrumented with pressure sensors across a single tube at both the condenser and evaporator sections

• Integrated an adjustable angle mount enabling experiments across multiple orientations and working fluids, with synchronized, temperature measurement across the tubes — generating large-scale experimental datasets for machine learning-driven design optimization and performance prediction

Oscillating heat pipes (OHPs) are passive, wickless two-phase cooling devices that move heat through the rapid, chaotic oscillation of liquid slugs and vapor plugs inside a sealed tube network. Because this flow behavior is inherently unstable and difficult to characterize, a deep-learning approach to automate its analysis has been developed.

Using a YOLOv11-based computer vision model, we trained a network to identify and track liquid-slug and vapor-plug movement directly from high-speed flow visualization images, achieving 90–96% detection accuracy across a large image dataset. This automated tracking extracts real-time positional data — information that would take enormous manual effort to obtain from thousands of frames — and feeds it, along with the thermo-physical properties of the working fluid, into a machine learning model that predicts the OHP's effective thermal conductivity.

This work bridges high-speed experimental diagnostics and data-driven modeling, offering a scalable way to understand and predict thermal performance in two-phase cooling systems without relying on computationally expensive physics-based simulations alone.

Research centers on the design and fabrication of micro-Oscillating Heat Pipes (OHPs) for advanced thermal management applications. Using cleanroom microfabrication techniques — including photolithography, DRIE etching, and silicon-glass bonding, we develop miniaturized heat pipe structures capable of efficient passive heat transfer at small scales. This work spans the full pipeline from cleanroom fabrication to experimental characterization, including pressure measurement and thermal performance testing.

Image analysis techniques were applied to study droplet size distribution during condensation on nanochannel-depleted liquid-infused surfaces (dep-nc-LIS). By tracking how drop populations evolve over time, surface deterioration was quantified and drop shedding behavior was linked to heat transfer performance — demonstrating that surfaces with smaller, more frequently departing drops significantly outperform conventional flat silicon surfaces.

In this work, a simple wet-bench based process is used to create a durable superhydrophobic surface, where a porous silica inverse opal layer is first fabricated on a cleaned glass substrate followed by infusion and depletion of silicone oil on the porous layer. Superhydrophobicity is achieved by coating the oil-infused porous layer with candle soot. The resulting surface maintains a water contact angle (WCA) greater than 150° and roll-off angle (ROA) less than 10°, even after undergoing various tests, including high speed water jet impact up to 8.6 m/s, heating the surface to 350oC and tape peeling test, submersion in acetone, ethanol, and saline pool for 10 days, immersion in acid and alkaline solutions for 48 hours, exposure to organic liquids (honey, chocolate syrup and soy sauce) as well as fine particulate matter (refined flour), highlighting its potential for real-world applications. Additionally, when immersed in water to a depth of 10 cm for 30 days, the superhydrophobic surface shows a shiny plastron layer without any degradation to its superhydrophobic behavior, thus affirming its robustness. The presence of this stable plastron layer enables the surface to remain dry even underwater, making it suitable for applications in aquatic industries.

Designing flat coatings with tailored wettability is critical for a wide range of engineering and scientific applications. Co-sputtering is a widely adopted technique for producing such coatings; however, it typically requires time-consuming and iterative tuning of deposition parameters through trial-and-error experiments. To streamline this design process, in this study, machine learning (ML) models are employed to predict the water contact angle (WCA) of co-sputtered coatings. During the experimental data generation, a novel flat coating composed of silicon (Si) and magnesium (Mg) was developed, exhibiting superhydrophilic behavior with a WCA below 5°. The coating demonstrated short- and long-term durability, retaining its superhydrophilic characteristics for up to 30 days under ambient laboratory conditions. Remarkably, even after one year, the surface could recover its superhydrophilic state following basic chemical cleaning. Among the machine learning models tested, random forest exhibited the highest prediction accuracy. This study highlights the potential of machine learning as an effective tool for guiding the design of co-sputtered coatings with targeted wettability, significantly reducing the time and cost associated with traditional experimental approaches. 

Planar plug nozzles being a potential and integral part of the propulsion systems, the current investigation is focused on assessing the effect of incorporating sidewall on the flow field and pressure distribution. In view of this, experiments have been carried out at different Nozzle Pressure ratios (NPR) corresponds to the overexpanded nozzle operating conditions. Initial experiments have been conducted to obtain the steady pressure distribution along the centerline and sideline in the presence of the sidewall and without sidewall. In order to have a better understanding of the flow features, Schlieren and oil flow visualization is performed. Oil flow visualization, in conjunction with steady pressure distribution, has been analyzed successfully to identify the dominant flow feature and their locations. This exploration provides evidence of the presence of a recirculation bubble whose size varies depending on the NPR as well as the presence of the wall. Furthermore, Schlieren flow visualization reveals the flow as unsteady, thereby demanding detailed unsteady analysis for the flow field. Therefore, unsteady pressure measurements both along the centerline and sideline have also been done and further processed to obtain its frequency spectrum. Spectral analysis of the unsteady pressure signals revealed a clear difference between the attached and the separated flow. In addition, other flow features with a distinct temporal mode associated with and without lateral confinement are observed. The intensity of low-frequency unsteadiness is comparatively reduced in the absence of lateral confinement; however, on the contrary, the interaction region is fairly reduced under lateral confinement.

Erosion is a severe and most common problem in boiler tubes of power plant boiler, aircraft engine, and steam and jet turbines. Erosion wear causes the loss of material from the component gradually, reduces the performance and thus, finally leads to failure of component. Coating of the boiler steel (substrate) is one of the effective methods to reduce the erosion wear. The major aim in this work is to use high pressure cold spray technology to coat the boiler steel. Cold spray is a promising technology for depositing protective coatings. In this project work of boiler steel, two coating powders was coated over T11 boiler steel substrate by cold spray coating technology. The T11 boiler steel, both coated and uncoated was subjected to jet erosion test at high temperature at an impact angle of 30 & 90 along with SEM, EDS and XRD characterization. Erosion rate is observed more for impact angle of 30 than 90. It is observed that the coated specimen showed better wear resistance than uncoated ones