Research Samples

Student:  Nicholas Chen
Professor/Sponsor:  Professor Omer Savas
Mentor:  Daniel Grieb
Research Project Title:  Skin Friction Drag on a Body with Hydrophobic Coating and Microchannels




Student: Nicholas Chen
Professor/Sponsor: Professor Omer Savas
Mentor: Rachael Hager
Research Project Title: Phase Doppler Anemometry and Particle Image Velocimetry


The ability to measure droplet size and velocity has always been an area of importance in fluid dynamics. Specifically, this experiment studies the formation and behavior of water droplets subject to turbulent flow. Two primary laser techniques were used to investigate these properties: Phase Doppler Anemometry (PDA) and Particle Image Velocimetry (PIV). The use of these techniques has the benefit of being non-intrusive to the particles.


PDA uses the interference of two monochromatic light sources to create a fringe pattern. When the droplet passes through the fringe pattern, a Doppler burst can be detected by photodetectors. This method is within the Eulerian frame of reference, or that the fluid flow properties are measured within a spatial reference instead of a temporal reference.


Of the two techniques, the primary technique used was PIV. This technique involves using lasers and lenses to create a light sheet which illuminates tracer particles within a plane and cameras capable of capturing particles to obtain high resolution images within small time frames. This method requires high particle density in each frame to fully capture flow properties. These particles must also be small enough to faithfully follow the flow fields instead of falling due to gravity. The captured image pairs are discretized into interrogation windows and then cross correlated to obtain displacement vectors for each window. Knowing the time steps between each image pair, the velocity of each particle can then be calculated. This then creates a velocity field where flow properties can be studied.


Student:  Glen Chou

Professor/Sponsor:  Professor Reza Alam
Research Project Title:  Acousto-optic modulation of brain waves



A thorough understanding of neural circuits remains an elusive goal in neuroscience due to the fundamental difficulties of monitoring and modulating neural activity in all regions of the brain. With this knowledge, the scientific community would experience a revolution in the design of prosthetic devices and the treatment of diseases caused by the central nervous system. The primary obstacle in the development of techniques in neural modulation is the fact that it is difficult to design a minimally invasive method for in-vivo experiments. In other words, it is difficult to get the light into the right areas of the brain without damaging the brain itself. To rectify this problem, our team is experimenting with acoustic wave-guides, using ultrasonic transducers, that will allow us to bend the light waves through the brain in a way such that brain damage is averted.


The goal of our current experiments is to prove that we are capable of bending light in arbitrary directions with an array of ultrasonic transducers. The transducer is placed in a small tank and is meant to generate pressure waves in the tank, which are then imaged using a camera. The laser, which is our light source, is shot into the wave tank and is aimed at the center of the transducer. The transducer’s waves then are meant to bend the light waves, causing the light to bend into a pattern of concentric circles (a Bessel function). Much of the research over the past few months involved trying to improve our image quality by creating a confocal microscope setup that would allow us to filter out noisy light rays. Simultaneously, we needed to characterize the electronic properties of the ultrasonic transducers and how their impedances depend on thickness. A major focus of the research this semester was in making multi-sectioned ultrasonic transducers and developing the circuits needed to drive them, which involves a phase delay array. In particular, the much of the time was spent designing a programmable circuit. When each segment of the multi-part transducer is driven by a voltage source with a different phase delay, the main lobe of the Bessel function can move and hence the laser beam can be steered arbitrarily.


Student: Sheyda Demooei

Professor/Sponsor: Professor Reza Alam
Research Project Title: The Bouncing Droplet


The bouncing droplet project studies the different phenomena surrounding the movements of small silicon oil droplets on a vibrating bath of silicon oil. The vibrating oil surface causes the oil droplet to bounce off of the wave and the air pocket between the oil bath and the droplet and move around the bath of oil. In this paper, we study the single slit experiment and the phenomena surrounding the recreation of this experiment with silicon oil. We try to recreate an exact replica of the single slit experiment using oil and obtain similar results. In addition, we study the affect of different parameters such as airflow and change in input frequency on the motion of the droplet and the single slit experiment. This paper discusses the methods and measures used in reproducing this experiment, the difficulties and mistakes and the results following the experiment.


Student: Kjell Ekman
Professor/Sponsor: Professor Liwei Lin
Mentor: Dr. Ryan Sochol
Research Project Title: Novel 3D Printed Microfluidic Components


Microfluidic components that are capable of autonomous "on-chip" operations are critical to the advancement of integrated fluidic circuitry for chemical and biological applications, including point-of-care (POC) molecular diagnostics and on-site chemical detection. Previously, researchers have utilized soft lithography to create microfluidic devices, but replacing this production method with high precision 3D printing leads to unprecedented freedom in design in all three axes. With the goal of producing diodes, transistors, and capacitors, we present a systematic method to designing, printing, cleaning, and prepping 3D printed devices to insure quality, reliability, and reproducibility. 3D printers represent a relatively new phenomenon on the market and have shown in our research to be unreliable in terms of uptime and therefore we have come up with a set of design guidelines to make optimal use of the 3D printer. The printing process used to create our diodes, transistors, and capacitors is limited by the resolution for printed features, strength of material, and support material abstraction. The combination of these three elements has meant that so far we have not been able to get significant results from our diodes and transistors but we have created a redesign of the internal structure that we hypothesize will lead to better real world performance.


Student:  Jack Guo
Professor/Sponsor:  Professor Mohammad-Reza Alam
Mentor:  Marcus Lehmann
Research Project Title:  Wave Carpet

Ocean Engineering/Fluid Dynamics






Ocean wave energy is a source of extremely viable but untapped renewable energy, capable of providing “10 % of global energy needs” [1], equivalent to $50 billion USD annually. The current global wave energy installed capacity is well below that of wind, biomass, and even geothermal; however, it is worth noting that wave energy potential has only barely been scratched. Apart from benefits such as proximity to demand sources, one main reason that wave energy is so attractive is its density of energy. As seen in the conceptual comparison portrayed in Figure 1, the energy harvested solar power incident on an area the size of a regulation soccer field is equivalent to the wave power of coastal waves incident on two to three meters of the Californian coastline.


In an effort to harness wave energy, CalWave and TAFLab are developing a “Wave Carpet” energy absorption device. As a qualified team in the Department of Energy’s Wave Energy Prize, CalWave derived design inspiration from the lab’s previous flexible carpet design. The current design has evolved to use a large rigid absorber board, mounted on top of the other components, to draw in the energy of incoming coastal waves.  In operation, the absorber board is displaced in heave motion by incoming waves, causing a displacement of the absorber board relative to a bottom “float” platform. Subsequently, the power takeoff (PTO) units’ hydraulic cylinders, which are sandwiched between the absorber board and float platform (as shown in Figure 2), use this energy of displacement to generate power. The entire system, which is submerged beneath the mean water level, is secured to the ocean seabed by a mooring system for safe operation.


This Fall 2015 semester, I was involved in research with this wave carpet project. My responsibilities involved manufacturing foam blocks to test buoyancy of the absorber plate and floating platform, experimentally optimizing the damper systems of the PTO units, and analytically researching an optimal design for the mooring system.




Manufacturing the foam required following accepted machine shop best practices to strive for consistently-sized blocks. A vertical band saw was used to smooth off faces; datum planes were successively established so that consistently sized blocks within a certain tolerance could be produced.


To analyze and optimize the dampers in the PTO units, a variety of different piston/cylinder combinations were machined in the UC Berkeley Mechanical Engineering Student Shop. These were then taken to the O’Brien Hall wave tank facilities, where I had constructed a test stand. Using a linear actuator coupled with LabView control software and Matlab processing, the other experimenters and I set up dampers in the test stand to determine damping values of the piston/cylinder system under different settings.


My mooring analysis task required an in-depth look into the optimal diameter, stiffness, design, and cost of the mooring line as well as the design of a seabed-attaching anchor. Mooring lines not only counteract the forces encountered in the system’s operation, but also safely secure the system to the seafloor and prevent dangerously high displacements in all potential wave conditions. As a result, an effective mooring system is extremely important for the Wave Carpet system. Using industry standards and mathematical analysis, I found preliminary results that also opened the door into further investigation that, hopefully, would also consider more complex methods.


Results and Conclusions


After the foam blocks were manufactured, they were incorporated into the 1:67 scale model for current testing as well as into the 1:50 scale model for competing in the Wave Energy Prize. After a wide array of dampers were manufactured and tested, the team settled on a design and size to incorporate into the 1:50 scale-model design’s PTO system.


From the mooring analysis, I found that in general, required stiffness decreases as the angle of the mooring cables from the horizontal increases. I determined that our team’s proposed design is within production feasibility, given current technological capabilities and offshore industry standards. The mooring system also satisfies resonance safety concerns. This preliminary analysis will hopefully open the door for further investigation, such as into the topics of vortex-induced vibrations (VIVs), drag viscous effects, and absorber plate resonance.



[1] Alam, Mohammad-Reza, “Wave Energy”, IRIS Magazine, Technology Avenue, Issue 1, Summer 2008, Pages 12-13.


Student: Ben Hightower
Professor/Sponsor: Professor Stephen Morris
Research Project Title: Extracting Contact Angles from an Extended Meniscus


Extracting the contact angle from the meniscus of an evaporating, perfectly wetting system is a complex problem. The contact angle is defined by the slope of the constant curvature region of the height profile extrapolated to H=0, but the identification of that region through the typical method of differentiation produces an abundance of noise. Nonetheless, the contact angle can yield great insight into the computation of the heat flow and the dependence upon the temperature difference and material properties of the system. The modified set of second order nonlinear differential equations is formulated into a boundary value problem representing the transport model, and is solved numerically in MATLAB via the shooting method. Lastly, the problem is attempted by integrating in the opposite direction, though the usefulness of this approach remains unclear.


Student:  Eric Ibarra
Professor/Sponsor:  Professor Ömer Savaş
Mentor:  Onur Recep Bilgi
Research Project Title: Experimental Investigation Of The Effects Of End-Tabs On The Thrust And Vortex Wake Of Marine Propellers



Student:  Dana Lynn Lansigan
Professor/Sponsor:  Professor Phillip Marcus
Mentor:  Max Jiang
Research Project Title:  Using a CNN and Fourier Transform Parameterization Technique to Predict Airfoil Force Coefficients



The convolutional neural network (CNN) is a powerful tool used in complex learning tasks. Recent work has shown that the CNN, most commonly used in image classification tasks, can also be applied to engineering design problems, including airfoil design. One challenge in implementing the CNN for airfoils is choosing a parameterization technique that sufficiently captures geometric information at the airfoil's boundary, where the most important physical phenomena happen. The goal of this project is to implement a Fourier transform parameterization technique and train a CNN to predict force coefficients on two dimensional airfoils. The CNN in its current state can achieve high r2 scores, but improvements to the network architecture can lead to more accurate results. The results are proof of concept that the CNN coupled with the proposed Fourier transform parameterization can produce high prediction accuracies.



Student:  Benjamin Lei         
Professor/Sponsor:  Professor Omer Savas
Mentor:  Rachael Hager
Research Project Title:  Phase Doppler Anemometry and Droplet size



During the course of the semester I worked with Rachael Hager in the Fluid Mechanics Laboratory and helped her with her principal research project which used Phase Doppler Anemometry (or PDA) to measure the size and velocity of droplets.  The experimental setup for PDA involved lasers a DAQ and two photodetectors as well as a small vial of liquid with particles.  In assisting her with her project, I created a signal amplifier and worked with another undergraduate to modify the source code of a data acquisition device (DAQ).  The reason for why I needed to create an amplifier was because the signals coming from the photodetectors were indistinguishable from noise (too small).  Then I learned C from scratch and worked with another undergraduate to modify the source code from the DAQ because we wanted to see how the data changed over time as opposed to getting a set of data for one period of time.


Student:  Parisa Lotfi

Professor/Sponsor:  Professor Ömer Savaş
Mentor:  Rachael Hager

Research Project Title:  Turbulence Effects on Cloud Droplet Dynamics and Growth


During the Fall semester, I have been collaborating with a PhD researcher in the experimental fluid dynamics laboratory. Together, we are experimentally investigating the effects of turbulence on cloud droplet dynamics and growth, which has applications in cloud seeding and is used in mitigating drought. Thus, we used several different experimental techniques to measure the velocity and diameter of the particles.


During the time of my research I collaborated with experimental set-up through designing and machining several parts. Once all the parts were ready, I helped with setting up the experiment and making sure laser and camera are aligned with the rest of the set-up before collecting data. Then we ran the experiment to measure velocity and size of the droplets using different experimental techniques. I performed Particle Image Velocimetry (PIV) to measure the velocity of the particles and Phase Doppler Anemometry (PDA) to measure the size of the particles overtime. I had to modify a code written in C++ and matlab that samples voltages from two photo detectors placed in different positions and calculates the phase shift of the particles. In order to receive accurate signal from the photo detectors and reject noise, I built an amplifier to amplify the signal and reject noise.

Student: Sikun Peng
Professor/Sponsor: Professor Liwei Lin
Mentor: Casey Glick
Project partners: Chengming Liu, My Chung
Research Project Title: Single Layer Microfluidic Transistor Via Optofluidic Lithography



My research is designing single-layer microfluidic Transistor, after successful designing of the current source last semester. Many fluidic devices require certain feedback in the fluidic system and a satisfying flow rate gain is desired in many bio experiments. The transistor is made with poly-dimethylsiloxane(PDMS) and undergoes soft-lithography process. 1% photoinitiator is introduced to the channel then a photomask is placed on the device and exposed to UV machine to make spring and piston in the channel. At the correct range of gate pressure, drain flow rate is proportional to source pressure. As gate pressure various, the piston in the channel moves and resistance in the channel changes. Based on the experiment, the transistor got a gain around 2.


Student:  Rachelle Rosano
Professor/Sponsor:  Professor Simo Makiharju
Mentor:  Daniel Grieb
Research Project Title:  Data processing and experiment setup for investigating drag reduction due to superhydrophobic coatings



Student: Jega Vigneshwaran

Professor/Sponsor: Professor Philip Marcus
Mentor: Chris Gebhart
Research Project Title: Examining Atmospheric Phenomena on Saturn using ACCIV


The purpose of this project is to study various atmospheric phenomena, such as the string of pearls and Saturn's hexagon, on the aforementioned planet by extracting velocity fields from photos that were taken of the planet by the satellite Cassini. Archaic methods of extracting the velocity fields from photographs involved using the human eye to locate features in photos of the planets, but the problem with this method, aside from the fact that it can be extremely time consuming, is that unless the time interval between the two photos is extremely small, the path the features travel will be curved, not straight, resulting in unacceptably large errors. ACCIV, or Advection Corrected Correlation Image Velocimetry, is an automated and more accurate method for extracting velocity fields from sets of images. The user inputs a set of parameters, such as correlation box size, stride, and search area, that have been optimized for the planet. The routine is then performed on a set of at least four images; the first and second image are separated by a "short" period of time (usually a few hours) and the third and fourth image are also separated by a "short" period of time; the two pairs of images are separated by a "long" period of time (usually 10-12 hours). A velocity field is extracted from the first two images and the last two images using CIV, or Correlation Image Velocimetry. An image from the first pair is then advected forward to the halfway time between the two pairs of images, while an image from the second pair is advected backwards. In the highly unlikely scenario the initial velocity fields are perfect, the two images at the halfway time should be identical. Correlations are then found between the two advected images before the images are advected back to find the corresponding tie-points on the initial images. Using this method, features from the Saturnian features can be more accurately tracked over longer periods of time.


Despite the high resolution photos captured by Cassini, the raw images cannot be immediately used for analysis. They must be first navigated and projected on a flat surface. There are two coordinate systems used in the projection of planetary images onto flat surfaces: planetographic and planetocentric coordi- nates. In a planetographic coordinate system, the latitude is defined as the angle between the equatorial plane and a line that is perpendicular to the body, while a planetocentric latitude is defined as the angle between the equatorial plane and a line connecting a point on the surface to the center of the planet. The photos used in the ACCIV analysis were already navigated and projected; ACCIV will not work with unnavigated or unprojected images.


Due to the large volume of photos taken by Cassini, there was a need to sort through the photos based on desired parameters. The photos are stored in a directory, with different subdirectories for each of the different filters, or combinations thereof, on the cameras. ACCIV, however, requires a set of images which show the same latitude and longitude over a period of time in order to extract the velocity fields. Thus, a code was written in order to sort through the image directory to find photos which contained the desired locations, defined by latitudinal and longitudinal bounds. Most of the photos were of the "entire" planet; the bounds of the image were from -180 degrees to 180 degrees longitude and -90 degrees to 90 degrees latitude. However, there are portions of the planet which are not visible to the satellite's cameras, so large swaths of pixels in the projected image are black, signifying a lack of information for that portion of the planet. The code first recursively searches through the specified directory in order to find all the HDF5 files (the images of Saturn are not saved in the conventional .jpg or .png formats, but instead in an HDF5 format). Then based on the geographic bounds and/or time bounds, the program searches through all the images to identify the ones displaying the desired region and copies the desired images to a separate folder.

Student: Aaron Wienkers
Professor/Sponsor: Professor Philip Marcus
Research Project Title: Approximating Stratified Unidirectional Shear Flows


Simulating the evolution of a shearing flow with continuous flow property profiles is very difficult analytically and becomes computationally expensive with increasing profile complexity. By first discretizing the velocity and vorticity space, and then the governing equations, the system can be approximated by piecewise functions, able to represent any arbitrary ow profile. These approximations increase solver efficiency, with little loss in accuracy if the interpolating functions are chosen strategically as to not introduce secondary instabilities into the system. Conditions for stability and consistency of piecewise linear density and vorticity profiles in unidirectional shearing flows are presented. With these limitations of using linear interpolation, error propagation through the system can be mitigated and an estimate on global error be found. The implications of utilizing increasingly higher order composite polynomial approximations to represent the flow profiles will be analyzed along with the effective regimes of each approximation. This work necessitates the development of a numerical technique for domain discretization of any continuous unidirectional shearing flow with effective methods to minimize residuals. These results will support a new and more efficient approach to produce the internal gravity wave dispersion relation for a continuously stratified fluid in an infinite domain by iteratively solving the eigensystem using discrete, constant density divisions. A solution via induction by discretizing the infinite domain while mitigating effects of vorticity in each subdomain was shown to exist under specific conditions explored here. This inductive approach can be applied to a range of problems where, although the startup computation cost is large, future efficiency is gained by discretizing the domain before, rather than after, solving the governing equations.

Student: Aaron Wienkers
Professor/Sponsor: Professor Philip Marcus
Mentors: Chris McKee and Richard Klein
Sub Area: Astrophysics and Turbulence Research Project Title: The Laminarization of Baroclinic Accretion Disks from Turbulent Core Collapse


Late protostellar accretion disks are often idealized as thin, Keplerian, and laminar in nature; however, many disk instabilities are not insensitive to the initial turbulence spectrum. One such mode of turbulence driving in protostellar disks is by anisotropic core accretion. We use the adaptive mesh refinement (AMR) code, Orion, to perform high- resolution simulations of solar mass star-forming molecular cloud cores located in massive star-forming regions. The turbulence and laminarization of the ensuing late protostellar and early protoplanetary disks are studied during periods of high mass-infall rates. We include self-gravity, use a baroclinic equation of state, and represent regions exceeding the maximum grid resolution with sink particles, accurately simulating Bondi accretion.


Presented is a preliminary report outlining the development of a baroclinic cooling prescription for global core collapse simulations forming T Tauri protostellar systems. Our model self-consistently treats the viscously heated disk equilibrium temperature and cooling time using the global disk properties. These results will be used to initialize a culminating study of baroclinic instabilities in protoplanetary disks.


Student:  Sean Luna

Professor/Sponsor:  Professor Reza Alam

Mentor:  Matthieu Laurent

Research Project Title:  Wave Carpet Optimization for Nonlinear Wave Regimes

Research Areas:  Energy Science and Technology, Fluids, Ocean Engineering



The Wave Carpet is a flexible underwater structure that generates energy as waves pass over it. It was conceptualized by M. Alam after he noticed that regions of coastline are protected from large waves; the muddy seafloor in these regions attenuates incoming waves which provides safe zones for small boats during storms. Replacing the seafloor with a flexible structure allows for energy capture from the waves for use in power generation, desalination, and other uses. This idea has evolved as it was implemented, particularly that the carpet is in the water column instead of the seafloor. This change introduces pressures on both the top and bottom of the carpet. A significant point of interest is modeling the forces experienced by the Wave Carpet to determine structural parameters and power output potential. We model the carpet as a rigid board to identify the maximum forces acting on the carpet. Previous studies analyzed the optimization of the carpet for the linear wave regime and found that waves with relatively small amplitudes are described well by linear wave theory. Deviations from this theory increase as the amplitude and period of the wave approach the Stokes or cnoidal regimes. Here we investigate the load exerted on the board by the water column as well as the pressure distribution on the board. We show that these factors deviate quickly from linear theory as we move to nonlinear wave regimes. These results provide improved optimization and prediction abilities for the Wave Carpet for the wide variety of waves that will be encountered in true environmental conditions.


Student:  Susanna Pesonen

Professor/Sponsor:  Professor Simo Makiharju

Mentor:  Monica Li

Research Project Title:  Engineering Design of Flow Conditioner



The purpose of this paper is to provide an overview of the design process of the flow conditioner that will be a part of a flow loop in the FLOW laboratory at UC Berkeley. In this flow loops multiphase flow will be researched and some potential areas of study include cavitation which poses many possibilities in a wide variety of topics. The design process had many considerations including economic feasibility and structural integrity. A rectangular design was created to meet all design specifications which were verified using finite element analysis and mathematical methods. Afterwards, then consultations were made with local machines shops about how plausible the design was. It was concluded that in order to ensure that the design would meet all tolerances when machines some critical design changes would need to be made.  A backup cylindrical design was also made which would be guaranteed to meet design requirements, but lack accessibility to the inside because it would not include a door.


Student:  Michael Rogers

Professor/Sponsor:  Professor Omer Savas

Mentor:  Rachael Hager

Research Project Title:  Phase Doppler Anemometry



The formation of clouds tends to take place when droplets of water are around 10-20 microns in diameter and are hypothesized to gradually increase as they collide with other droplets. The energy spectrum is hypothesized to decrease as diameter increases. One way of measuring the diameter of small particles is the use of Phase Doppler Anemometry (PDA) which is similar to Laser Doppler Velocimetry (LDV). It turns out that the geometry of the setup allows the diameter of small particles to be calculated with PDA if either the frequency or velocity with which particles pass through a small volume, is known. Velocity can be obtained from Particle Image Velocimetry (PIV), but finding the correct frequency is more challenging. Much of this semester was spent researching theoretical aspects of PDA (how PDA works, the “2π ; ambiguity,” etc.), collecting/analyzing large data sets, and refining the physical PDA system that was built in the fluid mechanics lab by Rachael Hager. Since PDA has never been performed at UC Berkeley, care needs to be taken to ensure that the PDA system is correctly measuring microspheres of known diameter so that microdroplets of water (unknown diameter) can be measured accurately. We first shortened and slightly modified the code so that it’s able to read data from the DAQ and process it faster. After we determined that the code was working properly, we turned to the pump system. We ended up redesigning it with new tubes and motors, but the decrease in diameter of the tubes caused particles to go too fast and there was cavitation bubbles present, which show up as if they’re spheres and make measurements ambiguous. We tried to use a motor to stir the particles up in lieu of a pump, but also faced the issue of cavitation bubbles, regardless of blade placement. Our current approach is to use a pendulum that swings back and forth over a small distance, so that each time it passes in front of the photodetector it refracts light. A clear, acrylic box filled with a gel that has the same index of refraction as acrylic should be able to refract light traveling at a known frequency, and after postprocessing we expect a normal distribution about the mean diameter.