Efficiency Beyond Computing: Exploring Heat Recovery in Data Centers for District Heating and Power Production

Semester project or Master project
Fall 2024

Context

In response to the growing demand for Data Centers (DCs) and the essential need for efficient computing solutions, this project delves into innovative approaches to address associated challenges. A key focal point is the large amount of low-grade heat generated by conventional air-cooled DCs. To optimise overall energy efficiency, contemporary cooling technologies advocate for the adoption of liquid cooling methods for chips. This project undertakes a comprehensive exploration with a specific focus on leveraging DC waste heat for both District Heating Networks (DHNs) and electricity production. The proposed approach involves implementing an Organic Rankine Cycle (ORC) and using the heat dissipated in its condenser as a heat source for heat pumps (HPs) used in DHNs. Within the scope of this project, the definition of the district will be the the EPFL campus in Lausanne.

This project addresses the multifaceted challenge of valorizing the waste heat generated by DCs. Specifically, DCs can be integrated into energy communities by providing the local district with heating services. Simultaneously, the exploration extends to the integration with ORCs, where the recovered heat is converted into electricity, which can be used to abate the electricity consumption of the heat pumps. By comprehensively understanding these aspects, the project aims to optimise energy efficiency within DCs, while also contributing to broader sustainability goals in district heating and electricity production.

Project/Tasks

  • Literature review and data acquisition. Study of alternatives for data center heat recovery.
  • Definition of the conventional (existing) system, and proposal of multi-energy systems and scenarios.
  • Model implementation. Optimisation of the overall energy system based on different objective functions (e.g. investment cost, annual operation cost, environmental impact).
  • Results generation for optimal solutions and comparison with the reference system.
  • Sensitivity analysis:
    • On the different temperature levels of the heat recovered from the heat source (data center);
    • On the degree of heating at the condenser, analysing its effect on both the ORC output and reduction in the heat pump energy consumption.

Skills

  • Interest in energy system design and optimisation.
  • Understanding of thermodynamics and modelling of energy conversion and energy storage technologies.
  • Programming skills in AMPL, Lua, Python or others are of advantage.
  • Results interpretation and report writing
  • Lectures:
    • ME-454 Modelling and Optimisation of Energy Systems
    • ME-409 Energy Conversion and Renewable Energy

Administrative

This project is part of a collaborative research project called “Heating Bits” that involves multiple labs from EPFL, including IPESE at EPFL-Sion. The project will be supervised by Sai Sudharshan Ravi (PhD student, IPESE) and Dr. Eduardo Pina (postdoc, IPESE). If interested, please send your CV, with a short motivation letter, to sai.ravi@epfl.ch.