This project will primarily be based at The University of Liverpool
A virtual engineering simulator delivers a key opportunity for the education of future experts as well as the essential future operators for new experimental facilities, while a zero-power experiment for advanced reactors (i.e. molten salt reactors, high-temperature gas-cooled reactors) is the door opener into a new, more sustainable and economic nuclear technology for the 21st century. The new generation of reactors designs offers improved safety, sustainability, efficiency, and cost. For example, molten salt reactors allow a highly integrated closed fuel cycle approach supporting the disruptive solution to operate directly on spent nuclear fuel.
To be successful, the UK needs to expand the number of available reactor physics specialists. The development of a virtual reactor environment gives future operators a testbed to create the understanding required for the safe and reliable operation of such a new experimental facility. This will in addition create a cutting-edge education opportunity for students to tackle the extremely high education demand in reactor physics to close a substantial gap on the way to the development of any kind of advanced reactors for the UK in the future. Machine learning (ML) methods can help to empower/revolutionize the simulator technology for nuclear reactors and enhance the experience for the users. In the current project, machine learning methods will be used for the creation and test of a zero-power reactor model based on operational data from a real experimental facility as well as physics-based models. The ML environment will allow to use the same simulator environment during the whole development cycle of a reactor technology by using different learning sets, dependent on the current development status of the programme.
The aim of this PhD project is to apply machine learning methods for the creation of the reactor model and test it against traditional modelling and simulation tools. This approach, if successful, will simplify and streamline the process of the simulator development since in the case of the ML approach a physical model can be used as well as later experimental data, can be used as learning set which will be required to build the model which then can be used for both cases. It will also unify the approach to the development of any zero-power simulator both future and existing.
GREENCDT
More detail about the ways in which we positively promote equality and diversity may be found in our Diversity and Equality of Opportunity Policy.
Continue with Facebook
