Research
Here is a summary of my research so far. Most of my publications can be found at ResearchGate. Feel free to reach out to me if you would like more information / are interested in collaboration.
Atomization, droplets and sprays
The process of disintegration of a coherent mass of liquid into a large population of droplets is commonly referred to as atomization. A large part of my graduate research at UW-Madison focused on developing a solid physical understanding of atomization. We studied the following aspects of atomization and sprays, and developed reduced order mathematical models to describe the essential physics.
1. High-speed, turbulent, primary atomization (RG)
In a nutshell, for the conditions relevant to diesel engines, we found that the mechanism of primary atomization is largely insensitive to surface tension and liquid viscosity effects, and is influenced by the gas-liquid density ratio. On the other hand, droplet formation is strongly influenced by fluid properties.
Downstream of the location of primary atomization, the flow is characterized by the motion of clouds of droplets -- a spray. We have studied sprays subjected to cross flowing air current. Sprays in crossflow (SIC) occur in applications including diesel engines, agricultural spraying, desuperheating / humidification in ducts etc. In one of our publications (also here) on the topic, we were able to categorize the sprays as those in strong or in weak crossflows. The regimes directly influence the overall topology, penetration and spread of the spray, and have important implications to applications where targeted spraying is essential.
The SIC work has been presented at ILASS (2011), and the presentation received the Simmons Award.
Air entrainment by plunging jets
When a stream of liquid plunges into a nominally quiescent pool of the same liquid, air bubbles can become entrained into the pool. The entrained bubbles can be transported in the pool, giving rise to a bubbly flow. Such bubble laden flows occur quite frequently in the nature as well as in industrial settings. The bubbly flow leads to the diffusion of gases into the liquid phase; some application where this is desirable include waste water treatment, gas-liquid reacting flows, fermentation etc. However, in a number of instances including the pouring of paints and liquid metals, the induction of gas bubbles is highly undesirable.
Our work on this topic was motivated by the need to understand the origins of entrainment of large volumes of air in the wake of water-jet-propelled ships. This situation is unique in that the plunging jets are impinge at very shallow inclinations -- and this leads to quite interesting, self sustaining ways by which large pockets of air (larger than the jet diameter itself) are formed nearly periodically.
Through detailed, validated computational simulations of the ensuing two-phase flow, we determined that the shallow impingement results in a stagnation point flow at the pool, causing the formation of a wide cavity in the pool. This cavity, subsequently, fills up due to gravity and deflects the jet in the process. Such deflection gives rise to the following stagnation flow event, and the process continues, giving the appearance of periodicity. We were able to characterize this periodicity purely in terms of Froude number. Such a periodicity does not occur in the case of jets impinging at steeper angles (> 20 degrees with respect to horizontal).
Two phase flow in stirred reactors
Stirred reactors are very commonly used in the chemical process industry to bring about homogenization and reaction between different fluids. Some of my research as an engineer at Dow relates to quantifying mixing under different conditions in stirred vessels. Some applications require the mixing vessels to be smooth on the process side. Under such a condition, the circular motion of the stirrer can lead to the formation of a depression in the gas-liquid interface within the vessel; this phenomenon is commonly referred to as 'vortex formation' in the mixing community. Often, it is essential to estimate the depth of this vortex, as too deep a vortex could interact with the stirrer and entrain gas bubbles, cause shaft vibrations, and reduce the life of shaft bearings and seals. On the other hand, too shallow a vortex is undesirable in applications requiring a draw-down of solids.
In this paper (RG), we describe the development of a new correlation to estimate the vortex depth over a wide range of Reynolds numbers (1e2 to 1e5) . The article has featured on the cover page of the 168th issue of the Chem. Eng. Sci. Journal.
Verification & validation of the vof solver, interFoam
Most of my multiphase computational research has been performed using the the volume of fluid (VOF) method implemented in the open-source CFD toolkit, OpenFOAM. We validated the volume of fluid solver (interFoam) throughly, and the key findings can be found here (RG).
As a side note, if you are interested in computational fluid dynamics (CFD), definitely take a look at OpenFOAM's official website, as well as this site, which is primarily run by some of OpenFOAM's very first developers. I have personally benefitted a lot by using this code, and I continue to use it and learn more about it.
Ongoing research engagements
I continue to work on interesting fluid dynamics problems relevant to the chemical process industry. The situations include two- and three-phase reactor designs, as well as biological systems.
Hobby project: Turbulence and its modeling
Turbulent flows are the rule, not an exception. We seem to know what turbulence means intuitively; yet, there is no rigorous way to distinguish between what we consider to be turbulent and not-turbulent. We often associate turbulence with extreme sensitivity to even small perturbations. Yet, there is something so robust about such flows that we are able to consistently use them to our benefit in our infinite industrial endeavors!
Understanding 'why a flow appears to be turbulent' is rapidly becoming my dearest project. Clearly, there is no overstating the value of statistical tools in describing such flows. However, what I am after is not the description, but a more mechanistic understanding. At this point, I am trying to formulate a reasonable set of questions to start my work. All I have right now is some notions...
I have chosen computing as my method of investigation as it provides me the necessary flexibility. Note, computing turbulent flows is no small challenge, as is evident from a vast collection of literature on modeling of such flows (DNS is still out of reach for flows of my interest). I find the Implicit Large Eddy Simulation (ILES) method to be elegant, and I have so far had fair success with it. However, as a part of this hobby project, I am testing the limits of this method as well.