Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Identification and control of a fuel cell system for peak shaving BEV demand
In this project we aim at perform a system identification of the fuel cell at ESI platform (PSI) and use the derived model in the context of stochastic MPC for peak shaving electric vehicles demand.
Keywords: system identification - stochastic model predictive control - peak shaving - hydrogen energy storage systems
In the course of the Swiss Energy Strategy 2050 the development of renewable energies has been fostered in order to achieve the carbon neutrality target. In order to allow for the integration of intermittent renewable sources, energy storage technologies are being developed. In this context, hydrogen represents a suitable candidate thanks to its scalability in storing energy for long periods of time. In the course of the recent years, PSI has developed the ESI platform in order to investigate the advantages of hydrogen energy storage systems (HESS). The ESI platform is an example of a power-to-gas-to-power line: renewable hydrogen is produced via water electrolysis in a PEM electrolyzer supplied by solar power and it is stored in pressurized vessels; when electricity is requested, the hydrogen is sent to a fuel cell where the chemical energy is converted back into power. In this project, the optimal control of hydrogen energy storage systems HESS is of interest: specifically, we aim at minimizing the averaged quarter-hourly peak power during the charging process of electric vehicles by making use of the HESS and controlling the fuel cell system (gas-top-power GtP process). Preliminary work has already been done on the electrolyzer where a system identification of the device had been performed and used in the context of model predictive control (MPC). With this project, we aim at closing the entire loop by performing a system identification of the fuel cell (i.e. efficiency map as function of pressure and temperature, states dynamics). We will then use the resulting model in a stochastic model-based controller for peak shaving the BEV fast charging in an optimal way. Finally, we will validate our newly-developed controller in the ESI platform via an experimental campaign.
In the course of the Swiss Energy Strategy 2050 the development of renewable energies has been fostered in order to achieve the carbon neutrality target. In order to allow for the integration of intermittent renewable sources, energy storage technologies are being developed. In this context, hydrogen represents a suitable candidate thanks to its scalability in storing energy for long periods of time. In the course of the recent years, PSI has developed the ESI platform in order to investigate the advantages of hydrogen energy storage systems (HESS). The ESI platform is an example of a power-to-gas-to-power line: renewable hydrogen is produced via water electrolysis in a PEM electrolyzer supplied by solar power and it is stored in pressurized vessels; when electricity is requested, the hydrogen is sent to a fuel cell where the chemical energy is converted back into power. In this project, the optimal control of hydrogen energy storage systems HESS is of interest: specifically, we aim at minimizing the averaged quarter-hourly peak power during the charging process of electric vehicles by making use of the HESS and controlling the fuel cell system (gas-top-power GtP process). Preliminary work has already been done on the electrolyzer where a system identification of the device had been performed and used in the context of model predictive control (MPC). With this project, we aim at closing the entire loop by performing a system identification of the fuel cell (i.e. efficiency map as function of pressure and temperature, states dynamics). We will then use the resulting model in a stochastic model-based controller for peak shaving the BEV fast charging in an optimal way. Finally, we will validate our newly-developed controller in the ESI platform via an experimental campaign.
1. The student will perform a literature review in the area of system identification and stochastic model predictive control.
2. The student will become familiar with the experimental setup at PSI and with the data available.
3. The student will perform the system identification of the fuel cell and will validate the resulting model on the real platform.
4. The student will develop a stochastic MPC scheme for the peak shaving minimization task where the loads are represented by the (stochastic) demands of electric-vehicles.
5. The student will verify the resulting controller in simulation and validate it on the real platform.
6. The student will write the report and prepare a presentation.
1. The student will perform a literature review in the area of system identification and stochastic model predictive control. 2. The student will become familiar with the experimental setup at PSI and with the data available. 3. The student will perform the system identification of the fuel cell and will validate the resulting model on the real platform. 4. The student will develop a stochastic MPC scheme for the peak shaving minimization task where the loads are represented by the (stochastic) demands of electric-vehicles. 5. The student will verify the resulting controller in simulation and validate it on the real platform. 6. The student will write the report and prepare a presentation.
Note:
The project will be supervised by PSI and Automatic Control Laboratory, ETH Zurich. The project is suitable both as semester project (where the emphasis will be placed exclusively on the system identification task) and as master thesis project (where both system identification and stochastic MPC will be explored). The project is specially suitable for students who enjoy facing practical challenges and working with real-world data and experimental setups.
Qualifications:
Interested Master students with solid mathematical foundations and good programming skills (either in Matlab or Python) are encouraged to apply. Familiarity with system identification and model predictive control are a prerequisite.
How to apply:
To apply send CV and updated transcripts of records (BSc and MSc) to:
• Christian Peter (christian.peter@psi.ch)
• Marta Fochesato (mfochesato@ethz.ch)
Note:
The project will be supervised by PSI and Automatic Control Laboratory, ETH Zurich. The project is suitable both as semester project (where the emphasis will be placed exclusively on the system identification task) and as master thesis project (where both system identification and stochastic MPC will be explored). The project is specially suitable for students who enjoy facing practical challenges and working with real-world data and experimental setups.
Qualifications:
Interested Master students with solid mathematical foundations and good programming skills (either in Matlab or Python) are encouraged to apply. Familiarity with system identification and model predictive control are a prerequisite.
How to apply:
To apply send CV and updated transcripts of records (BSc and MSc) to: