Research spotlight: Chiara Fusar Bassini applies machine learning to energy markets at MIT

What role can machine learning play in shaping the future of sustainable energy markets? This is a question Hertie School PhD researcher Chiara Fusar Bassini is trying to answer. With a diverse academic background, Chiara brings a unique perspective to the study of energy market dynamics. 

Currently on a research stay at MIT’s Laboratory for Information and Decision Systems (LIDS), she explores how robust machine learning methods can optimise energy systems while ensuring economic and environmental sustainability. In this interview, Chiara reflects on her experience at MIT, the challenges and opportunities of academic life in the US, and why a research stay abroad can be a valuable experience for scholars in her field.

Tell us a little about your academic background and the topic of your PhD research.

I have a mixed background, with a Bachelor of Economics and Computer Science and a Master of Scientific Computing, and I worked for a few years in the energy industry before going back to academia. My research focusses on applications of machine learning to energy markets. Energy markets are very complex because they are ruled both by economic rules and physical constraints. The demand for electricity, with its seasonal and daily fluctuations, allows for very high price volatility even within the span of a few hours. I am interested in understanding how we can leverage electricity data to draw learnings from current market structures and construct more resilient and sustainable electricity markets in the future. 

Can you elaborate on your experience at MIT? What department are you in, what are you working on, who are the people you work closely with, and how does it align with your ongoing research?

I work closely with my second supervisor, Prof. Priya Donti. Her research group is located within the Laboratory for Information and Decision Systems in the Electrical Engineering and Computer Science Department. The group works on robust and safe machine learning methods for energy systems. While many team members focus on network applications, I bring in the additional market component. That means answering questions such as: how do we make sure that we not only can run on energy grids efficiently, but also create the right incentives for more sustainable energy systems?

Given the volatility and complexity of energy markets, what challenges have you encountered in applying machine learning to these markets? How do you overcome them?

The main challenge is that different market designs have different drawbacks and hence require different machine learning methods. For example, in most European markets, electricity generators bid a price per megawatt for the amount they can produce the next day. The market follows a simple clearing rule, dispatching generators in ascending order of prices, but it does not account for grid constraints or startup costs. This can lead to costly changes in the actual power plant dispatch with respect to market outcomes: here, machine learning can help by reducing the need for such adjustments. In contrast, US markets integrate these constraints directly into the market clearing process, requiring a more complex optimal dispatch model to clear the market. Machine learning can help accelerate these models. Many of my colleagues specialise in this area.

Can you share what a typical day looks like for you?

Honestly, it may not sound glamorous, but research involves a decent number of hours staring at a laptop. I am an early morning person, so my day typically starts with a coffee and either coding, writing or reviewing literature. I typically have a few recurring meetings with my supervisors or with other PhD colleagues. I sometimes attend internal events as well, mostly organised by climate and energy-related groups such as the MIT Energy Initiative. In the evening many international student associations organise social and cultural gatherings on campus.

What would you say are the main differences between campus life and the research environments at Hertie and MIT?

The understanding of campus life is completely different here at MIT compared to the European universities I have attended so far. A student starts their day at campus in the early morning and does not leave it until late in the evening. The breadth of services offered, spanning from gyms to extracurriculars to childcare, is impressive. This brings people closer, because they spend more time with each other than they would otherwise, and it gives them a sense of community and purpose. On a more critical note, although the research community is very collaborative, there is also an unstated but ubiquitous pressure of competition, which can be hard on one’s mental health.

How have you been adapting to life in Boston and would you recommend a research stay abroad to your Hertie School peers?

I feel like this is not a one-size-fits-all question, because it depends on the stage of life one is in. A research stay can be meaningful, and I absolutely would recommend it, but only if it has a purpose. If you know what you want to get out of your research stay – be it a project, a collaboration, or a cultural experience – definitely go for it. In my case, I was very interested in the research at LIDS and wanted to understand what makes American research, and MIT in particular, so widely famous. Last but not least, it has been an excellent opportunity to reflect on my next steps.