Direct Simulation Monte Carlo (DSMC)
Direct Simulation Monte Carlo (DSMC) method uses probabilistic Monte Carlo simulation to solve the Boltzmann equation for rarefied gas flows, in which the mean free path of a molecule is of the same order (or greater) than a representative physical length scale. This method models fluid flows using simulation molecules that represent a large number of real molecules in a probabilistic simulation to solve the Boltzmann equation. Molecules are moved through a simulation of physical space in a realistic manner that is directly coupled to physical time such that unsteady flow characteristics can be modeled. Intermolecular collisions and molecule-surface collisions are calculated using probabilistic, phenomenological models. Common molecular models include the Hard Sphere model, the Variable Hard Sphere (VHS) model, and the Variable Soft Sphere (VSS) model. The fundamental assumption of the DSMC method is that the molecular movement and collision phases can be decoupled overtime periods that are smaller than the mean collision time. Currently, the DSMC method has been applied to the solution of flows ranging from the estimation of the Space Shuttle re-entry aerodynamics to the modeling micro-electro-mechanical systems (MEMS).
​
I was introduced to DSMC simulations in 2019. Since then, I have been part of several research projects, where my job was to conduct DSMC simulations in the open-source software OpenFoam. Some of my key research works are
​
1. Analyzed the flow field for a conventional converging-diverging nozzle.
2. Investigated the micro-scale thrust vectoring phenomenon for a nanosatellite.
3. Developed a meshing tool for preparing blockMesh script.
4. Merged deep learning with DSMC for flow behavior prediction to save computational time.
5. Developed a post-processing tool for flow properties calculation.
​
One of my DSMC related publications has been accepted in the journal (details on publications section). Currently, I am developing a deep learning tool compatible with both the internal and the external flow for flow behavior prediction. I intend to be an expert in micro-scale flow analysis and incorporate the DSMC method with data science and experimental research.
Deep Neural Network and Data Science
In recent years, we have witnessed several pioneering advancements in deep learning (ML) and neural network (NN). Typically, a neural network can be characterized as a series of algorithms that can recognize the underlying relationships between a set of data accurately through a process that mimics the way the human brain operates. The accuracy of the NN depends especially on the accuracy of the available training data. Training data are extracted from physically realistic models of a system or process with different degrees of complexity. The idea of coupling computational physics and NN is fairly new and bears great promise.​
​
The need of merging deep learning with DSMC is substantially realized as a very long run time is needed for DSMC simulations. However, using different time series forecasting algorithms, we can overcome the limitation and predict the flow properties in advance and save significant computational time. Since my department does not offer a course in machine learning, I learned the basics of machine learning through Coursera in 2020. Under the supervision of Dr. A. B. M. Toufique Hasan, I then undertook a project that uses time series analysis to predict the microscale flow properties. The work has passed a benchmark study test for both the internal and external flows with sufficient accuracy and is currently being tested for complex internal and external flows. We are expecting to save up to 30% computational time using a deep learning algorithm. We are also developing a research tool that will ease the process for other researchers to use the deep learning methodologies in their study. We are hoping to launch the research tool by March 2020.
Heat and Mass Transfer
​To understand the flow phenomenon and interpret its behavior, it is necessary to have a good grasp of the macro-scale flow physics. I'm very proficient in thermo-fluid system modeling, especially in the laminar and turbulent systems. I have published two journal papers and one conference paper on convective heat transfer for different thermo-fluid systems. In these studies, I have used COMSOL multiphysics to simulate different heat transfer configurations. I have also investigated the flow boiling and phase changing material (PCM) characteristics. I want to understand nanoscale thermal transport to extract its application in manufacturing applications. Both the macro and micro scale analysis can significantly help us study the physical phenomenon of thermal transport.