[Admissions] Yonsei Grad School Interview (Mechanical Engineering) - Oral Exam Questions & Timeline

It's been a while, but this time I want to talk about my graduate school admission: getting into the Department of Mechanical Engineering at Yonsei University Graduate School. I've pieced together the application timeline, the documents I submitted, and the actual oral (interview) questions I was asked.

Failing KAIST, and Moving to My Own Lab

As you may have seen in my earlier KAIST post, I studied and interviewed to get into KAIST, but ultimately didn't make it. So I ended up joining the lab at my own university, where I had already been working as an undergraduate researcher.

The lab I worked in was the "Turbulence Lab", which runs classical fluid dynamics simulations while also applying AI to solve fluid problems.

Getting into my own university's lab was much easier than the whole KAIST process. Having already studied all of Dynamics, Thermodynamics, Fluid Mechanics, and Solid Mechanics (parts 1 and 2 each), plus Engineering Mathematics 1 and 2, for the KAIST major exam definitely helped. Some of the subjects I had already taken as an undergraduate, and the professors sometimes asked about those very courses during the interview.

Application Timeline (Spring 2020 Admission)

The timeline back then went like this.

For grad school admission, I submitted documents and then sat for an interview. The interview took place in Engineering Building D; I waited on a chair on the 6th floor. There were three professors during the interview.

Documents and the Research Plan

When filling out the application, I had to write a Statement of Purpose and Research Plan. The prompts (for Spring 2020 admission) were these three:

1. Self-introduction
2. Motivation for applying
3. Study and research plan

The documents I submitted included my TOEIC score, transcript, award history, and expected-graduation certificate, all bundled into a single PDF.

The Oral Exam (Interview) Process and Major Questions

After the oral exam candidates were announced, I took the oral exam. The process went like this.

Here's the setup: all of the Mechanical Engineering professors were split into groups of three, each in a separate lecture room, and the students were distributed among the rooms. I sat on a chair outside my assigned room, waited my turn, and went in when it was my turn. Which room you were assigned to, and therefore which professors you faced, made the questions entirely a matter of luck.

Once inside, they first asked which lab I had contacted, then asked several questions about my major.

Looking back now, I don't remember every detail, but let me recall a few. A self-introduction is a given. Below are some of the questions on the major subjects.

These are mostly basic concepts you learn early in Fluid Mechanics and Thermodynamics, so if you studied the fundamentals properly as an undergraduate, you can answer them. Here's how I answered (or should have answered).

Streamline, Pathline, Streakline

• Streamline: a line that is always tangent to the fluid velocity vector at a given instant. Fluid particles do not cross a streamline at that moment. It's the most commonly used of the three.
• Pathline (particle trajectory): the actual path a single fluid particle travels. Think of the trail left by a single drop of ink.
• Streakline: the line connecting all fluid particles that have passed through the same location. In an experiment, the line you see when you continuously inject dye at one point is the streakline.

It's a nice touch to add that in steady flow all three coincide, while in unsteady flow they differ.

Intensive Property, Extensive Property

• Intensive property: a property independent of the amount of matter. Temperature, pressure, and density are typical; split the system in half and the value stays the same.
• Extensive property: a property proportional to the amount of matter. Mass, volume, internal energy, and entropy are examples; split the system in half and the value halves too.
• Note: dividing an extensive property by mass gives an intensive property (specific volume, specific enthalpy, and so on).

Dominant Pole and Second-Order Approximation

• Dominant pole: the pole closest to the imaginary (jω) axis. Because it has the smallest real-part magnitude, it decays the slowest and effectively dominates the transient response.
• Poles far to the left decay quickly and contribute almost nothing to the response. A common rule of thumb is that a pole whose real part is at least 5 times farther out (5 to 10 times, depending on the source) can be neglected.
• So even a high-order system can be approximated as a second-order system by keeping only the dominant poles (usually a complex-conjugate pair). That's the "second-order approximation."

And the "there are 5 poles, so what order is the system?" question was a trap. Here's how to answer:

• A system's order equals the number of poles. Five poles means it is genuinely a fifth-order system.
• But if two of them (a dominant complex-conjugate pair) dominate the response and the other three are far enough away, only the behavior can be approximated as second-order; the order itself does not become second.
• In other words, distinguish between order (the number of poles) and approximate order (based on behavior).

You should be comfortable with all the concepts considered fundamental in Mechanical Engineering. I don't know exactly what will come up, but questions from the four core mechanics, or from Dynamics, Vibrations, and Heat Transfer, are all fair game.

In Thermodynamics, going a bit further, you might get the precise definitions of enthalpy and entropy; in Fluid Mechanics, intuition for Couette flow and Bernoulli's equation, and even the Navier-Stokes equations. Nothing would be surprising.

The final admission screen for Yonsei University Graduate School's Master's program in Mechanical Engineering (Spring 2020). Personal information (application number, name, student ID) is mosaicked.

Having already tasted defeat at KAIST and gone through a rough summer, getting into my own university's grad school honestly wasn't a huge thrill.

Choosing a Lab and Research Topic (Controlling Fluids with Reinforcement Learning)

At this lab, I had to decide whether to continue the fine-dust time-series prediction model I had been working on as an undergraduate researcher, or take on something new.

I liked the AI research I had done as an undergraduate researcher in 2018-2019, and I had a vague idea of actually applying it to fluid mechanics to control something. Or to design geometric structures that reduce drag; that kind of research appealed to me.

I had also learned about autonomous driving via reinforcement learning in a third-year undergraduate course (though I couldn't fully understand it at the time), and controlling fluids with reinforcement learning seemed really cool.

In my next post, I'll continue with how graduate school shaped my life and my current job, and I'll organize the detailed concepts there as well.