Reduce, by 25–33% within two years, the carbon footprint of EPFL generated by the meals served on campus by combining internal expertise in Data Science, Food service industry and Material flow analysis.

Food is one of three main levers (alongside transportation and buildings) to reduce resource use as well as climate and biodiversity impacts on university campuses. In 2018, it was responsible for 17% of EPFL’s total carbon footprint, with 1.5 million meals served per year and a current average carbon footprint of 4.7 kg CO2e per kg of food.

MegaBites project aims to improve the sustainability of EPFL’s food system using data science. The project is structured in three tiers building upon one another:

To start with, comprehensive data about EPFL’s end-to-end food system (from ingredient suppliers to kitchens to customers to waste bins) will be collected via a worldwide unique measurement infrastructure put in place by RESCO since 2019. Data will be pooled together in a well-structured “data lake” and made available for analysis.

Based on the resulting data-driven insights (which will be integrated into a simulation tool for real-time diagnostics), we will identify and deploy maximally effective campus-wide interventions, considering how to:

  1. supply more sustainable ingredients,
  2. incentivize low-carbon meals,
  3. reduce food waste in kitchens and restaurants, and
  4. evaluate the potential for treating leftovers on site.

Identify and deploy effective campus-wide interventions to supply more sustainable ingredients, incentivize low-carbon meals, reduce food waste in kitchens and restaurants, and evaluate the potential for treating leftovers on site. Implementation of these interventions will reduce EPFL’s carbon footprint. Disseminating these interventions across academic networks has the potential to lead to important reductions in carbon footprint across the globe.

Prof. Robert WEST, Data Science Lab,

Bruno ROSSIGNOL, Head of Catering and Retail EPFL,

Dr Vincent MOREAU, Project Manager, Data Science Lab,