Distributed control for demand-side management and ancillary services in smart cities

Backround

Heating and cooling in buildings is responsible for about 10% of global greenhouse gas emissions. My prior work shows that by just using a clever algorithm in your thermostat, and a bit of data, you can reduce the energy required to heat and cool a building by 30% without sacrificing user comfort. Scaling this up would shave off 3% of GHGs worldwide, which is about 2x the emissions caused by the entire aerospace industry.

Buildings are also large consumers of electricity, and therefore utilities are interested in collaborating with buildings to reduce energy consumption, and to move that consumption around in time. This is refered to as flexibility - the ability to adapt your consumption in real time.

Objectives

1.    Develop simulations of individual buildings, electric grids, district heating grids, and interconnect them into a high-fidelity simulation of an entire city
2.    Design large-scale coordination strategies using distributed control theory, game theory, optimization, machine learning, etc.
3.    Fit these models/simulations to data collected from live hardware
4.    Run experiments of the coordination strategies on live hardware

Experiments

We will have access to a variety of data sources and the ability to conduct real experiments on live hardware.

Lillestrom Kommune

1.     Approximately 241 buildings
2.     84 V2G electric car chargers
3.     18 swimming pools

FME Solar

1.     A number of user partners in the building and PV industry sector, tbd

International partners

1.     UCSD Microgrid
2.     DHG at UBC Okanogan
3.     NEST building at Empa

DESSI at ITS

1.     On-campus microgrid, including PV systems and inverters, batteries, V2G chargers
2.     Energy consumption data from households

Projects

Distributed model-predictive control for coordinating buildings

Data-driven control for building flexibility

Buildings as virtual power plants

Large scale automation of modeling buildings from data

Mechanisms for smart grid -- building interaction

Digital twins for smart cities

Background Needed

All of the projects will require knowledge of coding, either in Python or Matlab for simulation, and C/C++ for hardware. The projects will have a communal git repository, and students will be expected to adhere to sound coding practices, participate in peer coding reviews, etc.

Some familiarity with control theory is a plus, but a lot of it can be learned during the project. Knowledge of at least one of: physics, power systems, or thermodynamics is necessary for all of the projects.