Abstract
Floating offshore wind farms rely on accurate models of the winds, waves, and currents. Prior to construction, these inform the platforms’ and turbines’ designs, and throughout the wind farm's life they are necessary to plan inspection and maintenance operations. This engineering internship will assess new wave modelling methods at proposed offshore wind farm sites in the Celtic Sea. Using data collected from these sites, the intern will be responsible for extending a machine learning wave forecasting methodology beyond averaged statistical parameters to operating with spectral data, providing a more complete wave forecast that can better guide engineering decisions.
Project Description:
This internship will apply novel environmental modelling techniques to challenges in the offshore wind sector. The intern will build this machine learning-based spectral forecasting model and validate it against both conventional physics-based forecasts and an existing machine learning-based averaged wave parameter forecast for areas of interest in the Celtic Sea. This internship will support ongoing research with Celtic Sea Power who will provide access to data, and the UK Met Office, with whom new methods are being developed. This is broken into five tasks:
Task 1: Training and data preparation
As part of the data preparation step, the intern will work closely with the supervisor and the University of Exeter Renewable Energy Group to familiarise themselves with data processing and wave spectra using Matlab and Python, and resource characterization from an offshore wind farm perspective. The objective is to both familiarise the intern with the models and measurements used in offshore wind farm design, and to process all raw data for use in latter stages of the internship. During this phase, material from Exeter’s Renewable Energy taught undergraduate and masters programmes will be made available to the intern to help accelerate their learning and understanding in this field. [Weeks 1-2]
Task 2: Wave spectra parameterisation
Once the data has been prepared during the first phase of the project, this phase of the project explores wave spectra and different methods to parameterise the wave spectra. The aim of this task is to identify suitable methods for parameterising wave spectra, and to identify the conditions where specific parameterisation schemes may be more appropriate by comparing these against the measured spectra. The machine learning framework (Task 3) is sensitive to the number of variables; it is therefore advantageous to use parameterisation schemes rather than empirical spectra. [Weeks 3-5]
Task 3: Machine learning model development
The main aim of this internship is to extend a machine learning wave forecasting model to operate with spectra. This task, therefore, builds this model using the data from Task 1 and the parameterisation strategy identified in Task 2. The work will be completed first, considering only a single location and then moving to spatial forecasts. [Weeks 4-8].
Task 4: Model comparison
Task 4 will compare the outputs of the model from Task 3 against both conventional physics-based forecasts and the existing machine learning model for wave parameters. The comparison process will quantify the accuracy of the model and identify specific environmental conditions that the model both does and does not capture well, using the data analysis training completed in Task 1. This task will also use engineering models to characterise how the forecasts impact loads on offshore structures in a preliminary study. [Weeks 8-10]
Task 5: Reporting
Task 5 will prepare a final report detailing the analysis undertaken as well as a summary presentation to be presented to both members of the University of Exeter Renewable Energy team, but also interested industry parties including Celtic Sea Power, SimplyBlue Energy, and the UK Met Office. [Weeks 11-12].
Preferred intern working pattern: This internship is based within the University of Exeter’s Renewable Energy Group based on the Penryn Campus in Cornwall. The internship would be desk based, however, is flexible to accommodate changes in the start dates and working pattern based on the intern’s circumstances. The university does not have core hours and there is flexibility for the intern and supervisor to select working hours that suit the intern and supervisor. If necessary, the role can be carried out remotely. There is a final dissemination event that will be held in person on the Penryn Campus involving several stakeholders for which the candidate would need to travel if not locally based.
Can the internship be carried out from home (remotely): Yes
Will remote working equipment be provided: Yes