This section presents some of the technical/research projects that I worked on and thought would be interesting for your perusal. Instead of technical depth, the focus is more on the problem formulation, solution approach, and generic learning that could be applied somewhere else as well. Interested readers would find the links helpful for further exploration.
Slurry Flows and Cellular Automata
As one of the research topics I explored physics-based cellular automata to simulate particle laded flow and slurry flows. Although this is still at an academic level, the approach gave me great learnings in numerical simulations. Some key learnigs to share are:
Coding CA particle solver in C++. This was a major undertaking and interested readers please take a look at the Appendix section of the thesis.
I combined the above particle solver with fluid solver in Ansys FLUENTĀ®. This was done through User Defined Functions (UDF) and was not trivial.
As you delve deeper to capture the physics of particle flows soon you will come across challenges to discover new CA rules and philosophical part :-)
The animation on the left is a CA-based simulation for one of the experimental cases discussed in the thesis.
Managing End-of-Life Systems in Semiconductor Manufacturing Using Health Index
Equipment at the end of functional life poses several challenges for manufacturing operations and long-term asset sustainability. This is even critical for semiconductor manufacturing where equipment upgrades are capital intensive with a longer return horizon. This demands an objective and quantifiable approach to manage and monitor the end-of-life health of the manufacturing systems.
This work presents a practical three-level approach that brings together engineering, operational, and supply chain factors under a single indicator. A visualization heatmap correlates equipment health to manufacturing impact for engineering and commercial decision-making. The approach is simple yet proven to be effective in fab environments. The proposed analysis applies in formulating system priorities, upgrade strategies, and capital expenditures. Brief guidance on health improvement strategies, application to new fab, and cost-based evaluation are presented for practical use.
Chemical Mechanical Planarization (CMP) in Semiconductor Manufacturing
Chemical Mechanical Planarization is a tribological process used in semiconductor manufacturing either to remove excess material from the wafer surface or create a flat surface for the next layer. The schematic on the left illustrates typical CMP equipment. Note the interdependencies among various components: platen, polisher head, polisher pad, pad conditioner, slurry flow, and wafer. This involves contact mechanics/dynamics, chemistry, and fluid flow; all occurring at micro to nanometer length scale.
Therefore, it is extremely difficult to account for an entire CMP physics in the numerical simulation. Most of the simulation work remains at an academic scale and commercial processes are developed experimentally. From an execution perspective, the challenge is not only to remove selective material but also to know where to halt the process. Plus a plethora of materials makes up the wafer surface that changes with each new generation of semiconductors.
To explore further :
Chemical Mechanical Planarization of Microelectronic Materials , 2008, Steigerwald et al.
Fundamentals of Semiconductor Manufacturing and Process Control, 2006, May and Spanos.
Chemical Mechanical Planarization: Slurry Chemistry, Materials, and Mechanisms, 2010, Krishnan et al.
Chemical-Mechanical Planarization of Semiconductor Materials, 2004, Oliver.
Probabilistic Approach for Equipment Reliability
Problem: For large and complex systems, such as power plants or heavy machinery, many features, events, and processes that control the failure modes are not known with accuracy. Additionally, these systems have interdependencies and redundancies. Such complex systems are not mass produces therefore their time history is limited for meaningful data analytics. Despite this inherent uncertainty you still need to quantify system performance over time. How will you proceed?
Solution Approach: In these scenarios, a probabilistic analysis may provide an implementable solution. You capture the uncertainties at each component in the form of probability functions. This may need subjective assessment. Then propagate this uncertainty through the system model that accounts for interdependencies and redundancies.
Learnings: See various case studies done using GoldSim, a Monte-Carlo based simulation software here:
Simulation Platform: GoldSim
Application of Virtualization to Preserve Legacy Systems
Problem: For a legacy computer running on Real-Time Operating Systems, (e.g., RTOS, QNX 4, 1998) hardware sustaining is a daunting task. Most likely you will not be able to find those Pentium PCs, network cards, and PCI slots. However, there is no other choice than to run this system because the application software cannot be rewritten. What would you do?
Solution Approach: This is where virtualization technology may help. Besides VMware and VM VirtualBox, the KVM (Kernel-based Virtual Machine) is a potential Linux-based option that proved effective. A machine emulator QEMU can interoperate with KVM to run virtual machines at near-native speed. Use qcow file format that grows as data is added, particularly useful for large disks. The challenges are steep to virtualize hardware such as older generation PCI slots or printer ports, but still doable up to a certain extent.
Online Resources (you can find many more!):
Energy Dissipation During Impact
Experimental:
These experiments investigated the effect of wall thickness on the energy dissipation through sphereāthin plate elastoplastic impacts and elucidates the underlying impact phenomena. Experiments were performed on 0.635 cm and 0.476 cm diameter (d) spheres of various materials impacting aluminum 6061 plates of different thicknesses (t) with low impact velocities up to 3.1 m/s.
Besides restitution coefficient, indentation measurements and numerical simulations are performed to gain a detailed understanding of the contact process and energy dissipation mechanism. As the āt/dā ratio decreases, a considerable amount of energy is dissipated into flexural vibrations leading to a significantly lower COR value.
Numerical Simulation:
Here, the normal elastoplastic impact between a sphere and a thin plate is analyzed using an explicit finite element method (FEM). The impact involves plastic deformation and flexural vibrations when combined resulting in significant energy dissipation. One way to quantify the energy dissipation is to employ the coefficient of restitution (COR).
It is observed that, in addition to material properties and impact parameters, the energy dissipation is strongly dependent on the ratio of plate thickness to sphere diameter. For the impact velocities commonly observed in granular systems (V= 5 m/s or less), it was determined that the energy lost to flexural vibrations can be neglected if the plate thickness is more than twice the sphere diameter. In this scenario, the mode of energy dissipated is primarily due to the plasticity effects.
Theoretical
This work presents an experimentally validated analytical model for low-velocity impact between a sphere and a plate. The model accounts for plastic deformation as well as flexural vibrations. The results are compared against newly carried out experiments. The model well captures the effect of plate thickness-to sphere diameter ratio, impact velocity, and material properties.
The modelās generalized framework allows consideration for various expressions of contact parameters, critical velocity, and residual indentation. Moreover, it can be easily incorporated into particle-based or discrete element modeling approaches for granular flows to evaluate the realtime coefficient of restitution as opposed to assuming the constant value beforehand.
Cellular Automata
During the research, Cellular Automata (CA) is one of the simplest and most fascinating ideas I have ever encountered. It is not only mathematical but philosophical as well. These are discrete (space and time) and abstract computational models represented by cellular grids where each cell has a unique state (like a cell on an excel spreadsheet). The states evolve based on neighboring cells and rules. These rules, are often simple but lead to intriguing patterns, for example, the one on left using Golly.
At the core, it comes down to finding the rules that can represent natureās non-linear dynamics. The question is how to find them, intuition? Interest readers should definitely go through the book āThe new kind of scienceā.
Helpful softwares to explore further (if you can code that is even better!):
Articles by Stephan Wolfram:
Cellular automata as models of complexity, Nature, 1984.
Statistical mechanics of cellular automata, Rev. Mod. Phys., 1983.
A New kind of Science, book.
Effect of Pre-stressing on Shape Memory Alloy
Shape memory alloys are unique category of alloy that remember their shape and are able to return to that shape even after deformation. Nitinol 60 (NiTi60) is is one of them, it is a nickelātitanium alloy that has a unique combination of properties such as high hardness, low elastic modulus, and superelasticity. Recent studies of static indentation tests on prestressed samples demonstrated a considerable load capacity increase, which motivated this experimental work. The coefficient of restitution (COR) is measured for two NiTi60 spheres of different diameters (0.635 and 1.270 cm).
This work shows that mechanically stressing NiTi60 spheres enables them to achieve a normal COR greater than 0.9 when colliding with NiTi60 landing. This is one of the highest ever reported COR values for self-mated crystalline metal alloys at moderate impact velocities. Consequently, it is believed that this material may be a potential candidate for use in applications requiring highly resilient and wear resistant metallic materials.
LoRaWAN and Internet of Things
The LoRaWANĀ® is a Low Power, Long Range, Wide Area (LPWA) networking protocol designed to wirelessly connect battery-operated āthingsā to the internet in networks. In recent years it is gained significant momentum; it is ideal for applications that send a small amount of data with low bit rates. It can transmit data at a longer range compared to technologies like WiFi, Bluetooth. This makes LoRaWAN well suited for sensors and actuators that operate in low power mode.
Some interesting applications have been so far are agriculture, asset tracking, animal conservation, etc. Some practical aspects to look at prior to implementation:
Availability of LoRaWAN sensors: see LoRa Alliance
Sensor battery consumption
Amount of data transmission
Network rage requirement (especially in dense areas)
Monitoring Dashboard development (it could be resource-intensive)
Although promising, there are limitation with LoRaWAN, below articles present critical review of this technology.
Personalized Dashboard and Data Visualization
If you are looking beyond your calendar, weather, and Todo apps, and would like to build your custom dashboard then definitely check out the MagicMirror project. It is open-source, free, and maintained by a big community. You can implement various configurable modules in various categories (i.e., utility, finance, weather, health, etc.). For hardware, Raspberry Pi or any Linux machine will work (good use for older computers!). To describe the entire set-up and configurations is somewhat beyond the scope of this paragraph, however, there are things I would like to point to:
Before starting, get familiar with basic Linux shell commands, installation, and the pm2 process manager.
Basics of HTML, CSS, and Java will be extremely helpful in customization
Familiarity with API and how it pulls data from web
Basics of time series databases such as Influxdb
After above, go through online videos and web resources (there are plenty)
A powerful personal dashboard can be achieved if Grafana or Kibana dashboards are embedded within MagicMirror or combination of Influxdb and Python. Then it becomes a versatile tool for things such as home automation, digital life, web monitoring, and so on.
Preliminary module suggestions: MMM-Carousel, iFrame, MMM-iFrame, MMM-SmartWebDisplay, MMM-Todoist, MMM-PLOW, MMM-CalendarExt2, MMM-DateCounter, MMM-Worldclock.
Influence of Tribological Coatings on Impact Damage of Bearing Materials
In this work, the effect of hard tribological coatings was studied in terms of mitigating impact damage between tungsten carbide spherical elements and two different types of steel substrates. The coatings included a hard, highly elastic Tungsten-incorporated diamond-like carbon (W-DLC) coating at two different thicknesses and a harder, less elastic Chromium Nitride coating. Impacts were created using a drop-test rig described herein and characterized in three ways:
A measure of the coefficient of restitution during impact,
Investigation of the impact site using an optical interferometer,
Fixed ion beam cross sections of select impacts for observation of subsurface damage within the coating and substrate.
It was found that hard coatings on softer substrates such as 440C steel were able to mitigate surface damage up to a certain impact speed, depending on the coating, but were unable to influence the coefficient of restitution. On harder substrates like 52100 alloy steel, the coatings were found to increase the coefficient of restitution, indicating a reduction in energy loss due to plastic deformation, and to reduce damage at each tested speed. These effects and their potential influence on bearing performance are discussed in regard to impact mechanics, surface metrology, and the material properties of the coating and substrate acquired by nanoindentation.