Extensive Calculations, Patient Professor

Assistant Professor Alexandra Komrakova's research can take thousands of hours. She researches and teaches numerical modelling, using formulas and equations to describe phenomena that happen in the real world. Specifically, Komrakova’s passion is the numerical simulation of liquid-liquid dynamics and interactions.

Liquid-liquid dispersions, or emulsions, are used in many everyday applications such as cosmetics, foods, pharmaceutical products, but also resource extraction, including the transportation of heavy oil and tailings treatment. Komrakova studies the fundamentals of emulsions in order to suggest changes to manufacturing and production methods that will improve the qualities and lifetimes of these products we all rely on.

Using mathematical equations, Dr. Komrakova develops numerical models that describe, as closely as possible, events in fluid systems. A good example is an oil and water mixture. When you agitate a tank of water by stirring it, for example, a turbulent flow is generated. Adding oil to the agitated water will change the dynamics of that flow. One reason is that oil can be several times more viscous than water and so it behaves differently than water. Eventually the oil will disperse throughout the water in the form of droplets of various shapes and sizes. Komrakova simulates these conditions to model the droplet size distribution at different points to understand how and when the system reaches dynamic equilibrium. Modeling even this simple system is challenging.

“The most difficult part,” she says, “is that there are so many uncertainties and difficulties in the real world. Liquid-liquid systems are very sensitive to impurities, which can dramatically change the properties of the liquids. It’s challenging to simulate a real system with 100% accuracy. I must critically assess all of my numerical results.”

Since numerical models involve assumptions and simplifications which introduces deviation from the real system, Komrakova has to work within acceptable margins of error and these are different for different systems.

“Acceptable margins of error are big questions in my research,” she says. “In some industries, the goal is to be within a 1% margin of error compared to experimental results. On the other hand, for extremely complex systems, a 50% margin of error is good enough. It depends on the system and the application.”

Predictions in hand, Komrakova then goes back to the real world system and compares her answers with what actually happens. If the system is a simple one and the margin of error is low, then Komrakova’s predictions can be very accurate and she uses these simple systems as benchmarks to develop predictions for more complex systems. Once she has developed a model for a specific process or interaction, she can use the model to improve the performance of the system.

“If I know that my model replicates reality, then I can use it to analyze the performance of the system when it operates under conditions that cannot be created experimentally in the lab,” she says. 

Komrakova relies on high-performance computing to develop finely detailed models of real world systems, where 1 mm3  is simulated as 10003  grid points in numerical space.

“My computer is my lab,” she says. “The simulations run on Compute Canada clusters, which is Canada’s advanced research computing network. My models can take months of calculations using 1000 CPUs and several terrabytes of RAM.”

As passionate as Komrakova is about liquid-liquid dispersions, she is more passionate about teaching. And the patience she needs for research results, translates to her teaching practice. 

“I love to see that light of understanding in students’ eyes,” she says. “I make myself available to students any time they need, for as long as they need. I’m happy to sit with a student for hours if they need help.”

In September 2017, Dr. Komrakova will experiment with a new classroom app called Poll Everywhere. The technology will allow her to run quick multiple-choice quizzes, with answers being returned to her anonymously but in real time. She plans to include “I do not understand” as one of the answers students can pick so she’ll know if the class is ready to move on.

Komrakova is always looking for enthusiastic new students, so if you’re interested in numerical simulations, you have found your newest best teacher.