Going into the Cloud to Study Renewable Energy Extraction from Ocean Waves
By Cole Freniere
I joined the Computational Multiphase Flow Research Group in the summer after my sophomore year. Led by Dr. Mehdi Raessi, the research group is primarily focused on Computational Fluid Dynamics (CFD) simulations of two immiscible fluids interacting with moving or stationary solid bodies. Specifically, I was assigned to a project funded by the National Science Foundation (CBET Grant No. 1236462), which involved simulations of Ocean Wave Energy Converters interacting with ocean waves. These Wave Energy Converters are complex to simulate, and require a large supercomputer to run for an extended period of time. How quickly the supercomputer solves the problem mainly depends on its hardware – processors, network cables, etc. There are several supercomputers available to us as university researchers, but we wanted to explore a new option – Cloud Computing. Amazon, whom we all know for its large online store, also offers computing resources in the Cloud, which customers can essentially “rent.” It is possible to build a supercomputer in the cloud, for a cost that depends on how powerful the hardware is, and how long we use the resources. The grant I received from the OUR provided the funds we needed to “benchmark” Amazon’s Cloud to see if it would be an economically feasible option for us. The outcome of the study was that Amazon’s Cloud offers a high amount of flexibility, and short term benefits, but over the long term, it is not an economically feasible alternative to university supercomputers. This is mainly because we need access to these resources continuously for long periods of time, which is not the Cloud’s strong-suit at this point. However, it seems more appealing for smaller, short term projects, for example: engineering consulting work. The cloud is also compelling because it offers so many different sets of hardware, which all are a low price to test, and enables exploration of the many different flavors of hardware available.
In this OUR supported project, I worked with Prof. Mehdi Raessi and his PhD student Ashish Pathak (both from the Mechanical Engineering Dept.) as well as Dr. Gaurav Khanna, Professor at the Physics Department and the Associate Director of the Center for Scientific Computing & Visualization Research. Dr. Khanna is well known for his Playstation supercomputer that is used for Black hole simulations. It was a pleasure collaborating with the experts in computational physics, and a great learning experience.
I presented the results from this OUR funded project at the American Physical Society, Division of Fluid Dynamics conference, which had more than 2,000 attendees; this was during my senior year, which is rare for an undergraduate student. In addition, I gave presentations at three other local research conferences. I was also able to publish my work* in the journal of Computing in Science and Engineering, co-published by the IEEE Computer Society and the American Institute of Physics.
Freniere watching the clock count down before he begins his timed presentation at the American Physical Society Division of Fluid Dynamics conference, 2016.
My OUR funded project was very fruitful for many reasons. It provided insight into an unexplored alternative to meet my research group’s supercomputing needs, and the Journal publication contributes to a specialized area of research. Also, the project enabled me to experience presenting at a research conference, and publishing a paper. Finally, it was a valuable experience because I learned something about a growing field of interest – High Performance Computing (HPC). This changed the way I view science and made me reflect on the capability of computer models. Additionally, it made me wonder: to what extent can we really simulate physical phenomena? Nowadays, the scope of simulations that scientists and engineers are implementing is incredible, and new advancements are being made all the time. I find scientific computing a compelling subject, and it is the main reason I decided to pursue a Master’s Degree in Mechanical Engineering.
This project was related to my research on Ocean Wave Energy Converters, because it introduced me to the field of High Performance Computing, and gave me an idea about how the simulation model performs on different types of supercomputers. Supercomputers come in many different flavors, and it is not always apparent which flavor is the best, because different algorithms require specialized hardware to run efficiently. For instance, data analytics and molecular dynamics models would require a completely different structure of supercomputer for optimal efficiency. When we get access to better hardware, we can do larger simulations which solve higher levels of complexity of the ocean wave motion as it interacts with the Wave Energy Converter. This can significantly increase insight into the physical problem. I am very excited to present the Wave Energy research at the 2016 American Physical Society conference in Portland, Oregon.
Computer simulation of an ocean wave energy converter. The device is a buoyant flap which pivots around a shaft on the ocean floor. Courtesy of Ashish Pathak.
As a mechanical engineering student, I was compelled by the subject material, and I was eager to get involved with undergraduate research. I also enrolled in the 5 year BSMS program, which enabled me to take graduate courses my senior year, which count towards both a bachelor’s and master’s degree. In my view, undergraduate research can be an excellent way to accelerate a graduate degree. Research is also interesting because during the course of conducting a research project, one always learns something new; it never really gets old and, above all, it is intellectually challenging. Another noteworthy issue is that everybody’s research trajectory is different. So, I can’t really tell you what it’s like to do research–you will need to see for yourself!
*Cole Freniere, Ashish Pathak, Mehdi Raessi, Gaurav Khanna, “The Feasibility of Amazon’s Cloud Computing Platform for Parallel, GPU-Accelerated, Multiphase-Flow Simulations,”Computing in Science & Engineering, vol. 18, no. , pp. 68-77, Sept.-Oct. 2016, doi:10.1109/MCSE.2016.94.