Credit: CC0 Public Domain
Although it is rare for a preteen to be fascinated by thermodynamics, those who are enthralled with the subject may consider themselves fortunate to have found MIT. Madhumitha Ravichandran is proof of this. Ravichandran, a Ph.D. student at Nuclear Science and Engineering (NSE), first encountered the laws and principles of thermodynamics while studying middle school in Chennai, India. She says, "They made perfect sense to me." "I was looking at my refrigerator at home and wondered if I could someday build energy systems using these same principles." It was that moment that it all began, and I haven't stopped thinking about it since.
Her doctoral supervisor, Professor Matteo Bucci (NSE Assistant), has been helping her to draw on her thermodynamics knowledge in her research. Ravichandran & Bucci are now gaining crucial insights into the "boiling crise", a problem that has plagued the energy sector for years.
Ravichandran was already well-prepared for this work when she arrived at MIT in 2017. She was an undergraduate at India’s Sastra University and pursued research on "two phase flows" which examined the transitions water makes between its liquid-and gaseous states. During an internship at the Bucci Lab in 2017, she continued her research on droplet evaporation, and other related phenomena. Ravichandran says that it was an eye-opening learning experience. "Back in India, only 2 to 3% of mechanical engineering students were female, and there was no woman on the faculty. It was my first encounter with social inequalities due to my gender. I experienced some difficulties, to be honest.
MIT provided a refreshing contrast. She says that MIT gave her a lot of freedom, which made her extremely happy. "I felt encouraged to explore my ideas and included in the group. She was doubly thrilled when she found out that she had been accepted to MIT's graduate programs mid-internship.
Her research as a Ph.D. candidate has been a very similar one. Bucci added an urgency to her work on heat transfer and boiling. The boiling crisis is affecting nuclear reactors as well as other types of power plants that use steam generation to drive turbines, and they are now studying it. Water is heated in a light-water nuclear reactor by fuel rods that have experienced nuclear fission. The water that circulates past the rods boils is the most efficient way to heat remove. Heat transfer will be greatly diminished if there are too many bubbles on the surface. This can engulf the fuel rods with a layer of water vapor. This not only reduces power generation but can also cause heat transfer to be reduced. To avoid the dreaded meltdown accident, fuel rods must always be cooled.
To provide a sufficient safety margin, nuclear plants are designed to operate at low power ratings. This is done in order to prevent such a scenario. Ravichandran thinks these standards are too cautious because people don't know what conditions will bring on the boiling crisis. She believes this affects nuclear power's economic viability at a time where we urgently need to have carbon-free energy sources. Ravichandran and the Bucci Lab's other researchers are beginning to fill in some of our knowledge gaps.
Initial experiments were conducted to find out how fast bubbles form when water touches a hot surface. They also determined how large the bubbles grow and how much the surface temperature changes. Ravichandran says that while a typical experiment took only two minutes, it took three weeks to find every bubble and track its growth.
Bucci and she have teamed up to implement a machine-learning approach, which is based on neural networks technology, in order to streamline the process. Neural networks can recognize patterns, even those that are associated with bubble nucleation. Ravichandran states that these networks are data-hungry. The more data they receive, the better they perform. The network was trained using experimental results regarding bubble formation on different surfaces. After that, the networks were tested on surfaces for the which the NSE researchers didn't have any data.
The team has now tested the output of machine learning models and is trying to get them to predict when the bubble crisis will happen. The ultimate goal of the system is to be fully autonomous and can predict the boiling crisis and show why it occurs. It will also automatically shut down lab equipment and stop experiments from getting too hot.
Ravichandran & Bucci made important theoretical advancements in the interim, which they present in a paper published by Applied Physics Letters. The nuclear engineering community was divided on whether the boiling crisis was caused by bubbles covering fuel rod surfaces or bubbles growing on top and extending beyond the surface. Ravichandran & Bucci concluded that the boiling crisis is a surface-level phenomenon. They also identified three factors that cause the boiling crisis. The first is the number of bubbles formed over a certain surface area. Second, the average size of the bubbles. The third factor is the sum of the bubble frequency (the amount of bubbles that form at a site in a given time) and the time required for a bubble reach its maximum size.
Ravichandran is pleased to have shed new light on the issue, but recognizes that there is still much to do. Although her research agenda can be overwhelming and time-consuming, Ravichandran never forgets her roots and the isolation she felt as an undergraduate engineer. On her own initiative, she has been mentoring Indian female engineers, offering both career guidance and research guidance.
Ravichandran said that sometimes, it feels like there was a reason why I had to go through these early hardships. "That's why I decided that I wanted to become an educator. She is also grateful for all the opportunities that have been opened to her since she arrived at MIT. She was awarded a MathWorks Engineering Fellowship in 2021-22. She says that it now feels like the only limitations on her are those she has placed on herself.
Explore more Infrared cameras, artificial intelligence and the Physics of Boiling
Additional information: Madhumitha Ravichandran et. al., Decrypting boiling crises through data-driven analysis of high-resolution Infrared thermometry measures, Applied Physics Letters (2021). Information from the Journal: Applied Physics Letters Madhumitha Ravichandran et. al., Decrypting boiling crisis using data-driven explorations of high-resolution Infrared thermometry measures, (2021). DOI: 10.1063/5.0048391
