Project 1. Developing a machine learning algorithm to extract information from a distributed optical fiber sensing response for fast hotspot detection in HTS applications
Arooj Akbar, Nicolò Riva, Bertrand Dutoit, Applied Superconductivity Group, BC.216, BC.224, INR.118
This project was carried out within the European Union project FastGrid which aimed at improving the REBCO tape architecture by increasing the limitation electric field with the overall goal of making a low length economically feasible Superconducting Fault Current Limiter (SFCL) device. The optimized REBCO conductor was used to wind pancakes for a high voltage DC SFCL module (≈ 1.5 kA – 50 kV) operated at 67 K.
SFCLs can be used to limit fault currents in both meshed DC and AC grids by transitioning from superconducting to resistive state, in the presence of high currents. While the device is theoretically a great way to protect grids, the inherent non-homogeneity of critical current along the superconductor length can lead to localized heating, called hotspots, and ultimately destruction of the SFCL device. At EPFL under the European Union project FastGrid, an extremely efficient Mach-Zehnder interferometer (MZI) based optical fiber sensing technique has been developed and patented that can detect even singular hotspots within 10 ms to protect SFCLs.
Instead of using common quench and temperature detection methods, a unique and fast sensing technique has been developed here at EPFL in collaboration with the Fiber Optic Group.
Optical fiber sensing for HTS devices health monitoring presents a lot of challenges pertaining to integration and sensitivity. Specifically for the case of MZI based sensing, the optical fiber setup shows sensitivity to both temperature and strain caused due to the hotspot in the sample. The technique is also extremely sensitive to external disturbances, which means it can sense disturbances that arise not because of the hotspot but due to environmental disturbance, Lorentz force caused due to the current, acoustic noise etc. This can lead to false alarms and a response that is not uniform. For this project, extensive data analysis are necessary to study the response of our optical fiber sensing technique. It is important to see if Machine Learning algorithms can work to eliminate false alarms and extract information even from a bad quality non-uniform response to raise the alarm for a hotspot. This algorithm has then to be optimized to achieve a fast performance in real time.
Experiments can be performed in parallel to wind different lengths (gradually increase lengths) of HTS tapes, to characterize the disturbance on the optical fiber with different length coils to understand when, why and how the response quality degrades.
Project 2. Using FEM to quantify the strain experienced by the optical fiber in distributed optical fiber sensing for hotspot detection in HTS applications
It is also important to study, by means of simulation, the thermal and strain transfer from the HTS tape to the optical fiber in the event of a hotspot. It is also interesting to make models to study different integration architectures for the optical fiber with the HTS tape. This project, therefore, also necessitates finite element modelling (COMSOL/MATLAB) of HTS tapes with the integrated optical fiber and by means of thermal, mechanical or magneto-thermal models. These models will be used to study the sensitivity of our technique and correlate the thermal and strain transfer in the optical fiber to the frequency and speed of the response.
These projects aims to carry out extensive experiments on HTS tapes and the SFCL pancake to study the technique and demonstrate its feasibilty for HTS applications. The project will give a unique opportunity to the participating student to work hands on in the superconductivity lab and interact with a team of the best European laboratories in the field, that were involved in the FastGrid project.