About the project
To help reduce the carbon emission of shipping, Silverstream Technologies have developed an air lubrication system shown to reduce fuel consumption by 5-8%. The system releases a carpet of bubbles along a ship’s hull, providing a separation between the hull and the water, reducing the frictional drag. The quality of the carpet determines the potential fuel savings.
This is determined by:
- the power used to pump the air into the water, creating the carpet
- the operating modes of the ship
- the environmental conditions it is sailing in
This forms a complex optimisation problem that is difficult to model. It’s also difficult to replicate realistic operating conditions through experiments. In this course, you will focus on using machine learning to help understand the performance of the system and its sensitivity to these different input variables.
It will take data from ships operating with the system and correlate the input power, operating mode and environmental conditions with either high or low performance. This will determine where improvements can be made.
This information will be passed on to experts in fluid dynamics to help inform the modelling and tests that will benefit our understanding of the system. In a second stage the knowledge will then be used to determine how to reduce the required data for accurate predictions using a fusion of data and to automate the learning in the system. This will reduce the expert knowledge required to update the system during operation, allowing an efficient update of a fleet of models.
You will work closely with the Marine and Maritime institute, in the Data-Centric Engineering Programme in The Alan Turing Institute, the UK’s national AI institute and the Applied Research Group at Silverstream Technologies.
