Technical Articles

Nuclear Reactor Modeling Through Distance Learning

By Christophe Demaziere, Chalmers University of Technology


To help meet the growing need for graduates with nuclear safety training, a short course on nuclear reactor modeling, Deterministic Modeling of Nuclear Systems, was created for master’s students, doctoral students, and postdoctoral researchers at Chalmers University of Technology. The course is given as a week-long workshop in which students model a sodium-cooled fast reactor in MATLAB®, taking neutron transport, flow transport, heat transfer, and their interdependencies into account. Students complete all in-session coding exercises with MATLAB Grader™ (Figure 1).

Figure 1. A coding exercise in MATLAB Grader for the short course on nuclear reactor modeling.

Figure 1. A coding exercise in MATLAB Grader for the short course on nuclear reactor modeling.

Until now, the workshop was taught using a flipped classroom model and a hybrid approach in which some students attended in-person at Chalmers while others attended remotely online. Due to the COVID-19 pandemic, the workshop will most likely be conducted entirely online in the future. Because we were already using MATLAB Grader in the hybrid version of the course and had experience with teaching remotely, the transition to a fully distance learning setup will be straightforward. This setup has just been successfully applied following the same principles for part of a master’s-level course on computational continuum physics, offered for the first time at Chalmers in the spring of 2020. Because of the COVID-19 outbreak, that part of the course was entirely given online.

Workshop Motivation and Setup

Close to 100 nuclear reactors are still in operation across Europe, accounting for more than 25% of all electrical power generation. At the same time, however, nuclear engineering programs are being phased out at European universities, and government agencies warn of a dangerous lack of scientists and engineers with expertise in nuclear safety. Deterministic Modeling of Nuclear Systems was created, in part, to provide as many students as possible with in-depth training in the modeling of nuclear reactors, and as a result, in nuclear safety, regardless of their ability to attend in-person at Chalmers.

Bringing Students Up to Speed

When we originally opened Deterministic Modeling of Nuclear Systems to distance learning, one of our first challenges was to provide material suitable for students with diverse educational backgrounds. We addressed this challenge by providing the students with self-paced learning activities to be completed before the start of the one-week long workshop. Thanks to this preparatory work, the “face to face” sessions (either onsite or online) could concentrate on more active forms of learning via specifically designed activities.

The preclass preparatory work focused on two aspects: providing the necessary theoretical knowledge of reactor modeling (conceptual understanding) and making sure that the students could focus on solving the MATLAB tasks during the face-to-face sessions (procedural understanding). Some students came to the class with programming experience, some with MATLAB experience, and others with neither. We encouraged students who were unfamiliar with MATLAB to complete the free, hands-on practice sessions available via MATLAB Onramp, including the sessions on vectors and matrices, importing data, array calculations, calling functions, and plotting data. Students who had access to Campus-Wide MATLAB Training through their university could also opt to complete the three-day, self-paced, online MATLAB Fundamentals course, which covers the same material in greater depth.

Fostering Active Learning

In the approach we developed, all students learn independently and in their own time before the actual workshop takes place, thanks to teaching materials made available on the web. The class sessions focus on activities that engage students in higher-order thinking, with a focus on developing and implementing algorithms in MATLAB.

For the preparatory activities focusing on conceptual understanding, students are required to complete assigned readings from my textbook on nuclear reactor modeling (written specifically for the workshop), watch short recorded lectures or webcasts, and answer online quizzes. Whereas the textbook presents the details of nuclear reactor modeling, the webcasts summarize the key concepts presented in the textbook, thus further enhancing the student’s conceptual understanding.

During class sessions, I briefly review the key concepts and answers to the quizzes. The students then work on implementing the MATLAB based reactor modeling algorithms presented in the textbook and webcasts. To help students converge on the right solution, I created coding templates and assessment tests for each interactive exercise in MATLAB Grader (Figure 2).

Figure 2. Plots of axial sodium velocity (left) and temperature distribution within a fuel rod (right) created for an assessment test in MATLAB Grader.

Figure 2. Plots of axial sodium velocity (left) and temperature distribution within a fuel rod (right) created for an assessment test in MATLAB Grader.

MATLAB Grader automatically grades the exercises and provides feedback, leaving me and my teaching assistant free to work one-on-one with students, answering their questions and providing guidance when needed. The web-based MATLAB Grader interface also ensures that all students, whether local or remote, have access to the same coding environment.

Many students use the desktop version of MATLAB as a complement to MATLAB Grader to explore concepts in greater depth outside the scheduled sessions. Students who do not have access to MATLAB through their university can take advantage of a 30-day free trial.

Learning Outcomes and Student Feedback

For most of the students, this was their first experience of a flipped classroom. The combination of the flipped pedagogy with the MATLAB Grader coding assignments represented a “disruption” in the way they were used to learning. Nevertheless, the students were extremely positive about the setup, irrespective of whether they followed the course onsite or remotely. In course surveys, 80% of the students had a very good impression of the course and 20% a good impression (the choices were “very good,” “good,” “somewhat not good,” and “not good at all”). Furthermore, 36% of the students believed they learned much better in this teaching setup and 32% somewhat better; only 4% believed they learned much better in a traditional format and 8% somewhat better in a traditional format, while 20% believed they learn equally well in this teaching setup and in a traditional format. 48% of the students rated the pedagogical approach as very good, and 52% as good (the possible choices were “very good,” “good,” “somewhat not good,” and “not good at all”). Finally, and most importantly, 76% of the students found the synchronous sessions very engaging and 16% somewhat engaging. Just 8% found the sessions not engaging at all (the possible choices were “very engaging,” “somewhat engaging,” “somewhat not engaging,” and “not engaging at all”).

Beyond the level of engagement in the synchronous sessions (which, in my opinion, is a direct measure of the efficacy of the approach), all onsite students successfully completed all assignments. For the remote and active students (the ones who actually worked on the assignments), the completion rate was 98%.

I should mention that a thorough analysis of student learning is ongoing for our other fully online course, Computational Continuum Physics, using data collected by monitoring their learning activities throughout the learning sequence (preclass activities in the forms of reading the textbook, watching the webcasts, and answering the online quizzes; in-class activities while working on the MATLAB Grader assignments; and post-class activities for completing the assignments). Early results indicate that the level of student engagement was extremely high, despite the fully online nature of that course, and that students who participated performed extremely well on the coding assignments.

Distance Learning: Future Plans and Advice for Colleagues

Having taught distance learning courses for several years, I am often asked by colleagues for advice on getting started and strategies for improving the learning experience. I encourage these colleagues, first and foremost, to develop a plan for teaching the course and to follow the plan as closely as possible, even if they are converting to distance learning while the course is running. Difficulties are to be expected, especially when both the instructor and the students are new to distance learning. I’ve found that sticking to the plan through these difficulties is a better strategy than abandoning it mid-course to try something new. 

I encourage instructors to consider factors such as the amount of preparation time available and how many students they will have. Remember that most students are as unused to online learning as instructors are to online teaching. Provide a structure for the course that is easy to understand and follow. Finally, and most importantly, find ways to interact with students, offer hands-on training sessions, and make the students active learners. The in-class sessions, during which active learning and student-teacher interactions occur, are key to improving student learning, whether onsite, online, or in a hybrid setup. An automated grading system such as MATLAB Grader offers substantial advantages in this regard: It gives the instructor time to work with the students individually while giving students an opportunity to engage in hands-on, active learning with immediate feedback. The different learning elements and their interrelation have to capitalize on each other and support student learning, thus explaining why having a strategy is essential.

About the Author

Christophe Demaziere is a professor in the department of physics at Chalmers University of Technology, where he leads the DREAM (Deterministic REactor Modelling) task force, a cross-disciplinary group with expertise in neutron transport, fluid dynamics, heat transfer, and numerical methods.

Published 2020

View Articles for Related Capabilities

View Articles for Related Industries