Fazel Khayatian

Urban Energy Systems Laboratory

Empa - Swiss Federal Laboratories for Materials Science and Technology

Uberlandstrasse 129, CH-8600, Dübendorf, Switzerland

Tel: + 41 58 765 11 22

  Google Scholar



Fazel is a Research Scientist at the Urban Energy Systems Laboratory of Empa and Lecturer for the Institute of Technology in Architecture at ETH Zurich. Prior to joining Empa, he was Assistant Professor of Building Physics at the University of Nottingham in Ningbo China (2018-2019). Fazel is an Architect by training (BArch) and an Engineer (MSc Architectural Engineering) who has developed a particular interest in data-driven modelling of the built environment. He has a PhD from Politecnico di Milano, focusing on the applications of machine learning for multi-scale energy audit.

Currently, he explores physics-based building energy modelling, uncertainty and randomness in energy systems, as well as the applications of machine learning in building energy analytics. He is also interested in trustworthy machine learning and its applications in building performance assessment.



  • Dynamic CO2 Emission Model of Cities

The city of Zurich, has adopted the targets of the 2000 Watt Society, which proposes an 82% reduction in emissions by 2050 compared to 2005 levels. However, tracking progress towards these reduction targets requires consistent, reliable, and timely information on CO2 concentrations and emissions. In this project, we integrate bottom-up and top-down CO2 modelling approaches. Using a detailed building model including occupancy and heating and cooling systems, we provide better approximations of CO2 sources; particularly oil, natural gas, biomass, and district systems. This is combined with additional datasets (e.g. human activities, traffic, industrial emissions, power production) to estimate, “bottom-up” local contributions by each economic sector. Measurements of atmospheric CO2 concentrations provide independent “top-down” information. This project is a collaboration between the Urban Energy Systems and the Air Pollution / Environmental Technology Labs at Empa.


  • Algorithmic Regulation and Control (ARC)

In a classical control loop, feedback is needed for a system to adjust its state and minimize its deviation from a given set point. In the building sector, such feedback mechanisms are generally weak and are rarely based on real measurements, resulting in well documented performance gaps throughout the whole life-cycle. Consequently, buildings are not constructed and operated in an optimal way. The aim of this project is to set up and evaluate high fidelity physics-based building energy models, to overcome the information gap. The project is particularly focused on improving the design and operation of HVAC controllers by using digital twins.


Alessandro Tell (ETH Zurich, D-ITET) Master Thesis: Defence against adversarial inference from smart home data. (2022)

Sven Gluaser (ETH Zurich, D-INFK) Master Thesis: Inference attacks on smart home data. (2022)

Doris Lima (ETH Zurich, iFA), Master Thesis: A GAN-based framework for data-driven optimization programs. (2021)

Jordan Mignan (ETH Zurich, D-MAVT), Master Thesis: Coupled vs. Decoupled: A study on Modelling District Heating Systems. (2020)

Alicia Lerbinger (ETH Zurich, D-MAVT), Semester project: A Python Module for Linking Energyhub and EnergyPlus Tools. (2020)


Open positions

Semester projects:

Master theses:

Currently, there are no open positions.

Students who are seeking an internship and intrigued about the following topics can send me an unsolicited email describing their background and interests:

  • Synthetic data projection;
  • Adversarial inference;
  • Label poisoning.