Multi-fidelity systems-level simulations for nuclear thermal hydraulics

Thermal hydraulics simulations at the plant level are vital for the understanding of reactor performance and safety. However, full-resolution 3D computational fluid dynamics (CFD) simulations at the plant scale are intractable, even with Reynolds Averaged approaches. CFD best practice grid convergence dictates lower-end grid spacing requirements. Meanwhile, the overall dimensions of the plant dictate the computational domain size. The ratio of plant size to grid spacing is typically several orders of magnitude greater than can be practically solved, even with modern high-performance computing on thousands of compute cores.

For this reason, system-level simulations employ one-dimensional (1D) models to represent the flow and heat transfer in plant sub-systems. Such simulations simplify the complex 3D geometry of the sub-systems into a network of interconnected 1D sub‑components. In the most part, this works well, and allows tractable simulations at the plant scale with modest computational resources. Hydraulic losses and heat transfer coefficients throughout the network are generally modelled using empirical correlations, with the overall solution arising from conservation laws.

While 1D models perform satisfactorily in most of the sub-systems, there are inevitable compromises in such analyses. In particular, 3D flow effects cannot be resolved with 1D models; examples are thermal stratification layers and buoyancy-driven annular flows in regions of intense pipe-wall heating or cooling, which require assumptions to be made in 1D models. It is also possible for the local flow conditions to lie outside of the range in which the empirical models underpinning the systems code are valid. Accordingly, in order to allow tractable yet more accurate system-level simulations of nuclear power plants, a hybrid multi-fidelity (1D with 3D CFD) approach is needed.

This PhD will develop a multi-fidelity model of a thermal hydraulics problem involving a flow loop, which intelligently combines 3D CFD only where needed with 1D system code simulations during runtime. Bypassing some of the challenges associated with a traditional boundary-based code coupling, we will investigate a novel hybrid 1D/3D approach. The entire domain will be discretised with both 3D CFD and 1D system code meshes, with one locally driving the other via source terms. The CFD mesh will initially be coarse everywhere. Predictions on such coarse meshes should not be expected to outperform the 1D model prediction, and hence the 1D model will drive the CFD via drift terms to enforce consistency between the two overlapping domains. However In regions of the loop pipework where the 1D model would be expected to perform poorly (regions where thermal stratification or strong 3D effects are expected) the CFD mesh will then be refined. The resolved CFD will then drive the systems code in these areas via local drift terms (sources) in the 1D governing equations to enforce consistency between the solutions. In regions where the 1D systems code predictions are expected to be reliable, the CFD mesh will remain coarse and local source terms in the CFD governing equations will be used to drive its solution to be consistent with that of the systems code. The identification of which local regions should be driven by which code will initially be done manually. With machine learning, the potential to automate the identification of which code should be the driver throughout the domain can be considered. A prototype of this volumetric coupling has been developed within the team. The PhD student will take this idea from a prototype to a mature and thoroughly tested method.

This project is co-funded by Rolls-Royce Submarines. The PhD student will work closely with Rolls-Royce staff to maximise the industrial relevance of the work, including via a short-term secondment. Due to the need for security clearance while on secondment, this project is open to home students only. 

Eligibility

Applicants should have, or expect to achieve, at least a 2.1 honours degree or a master’s (or equivalent) in a relevant science or engineering related discipline.

 How to apply

• Please complete the  application enquiry form – SATURN CDT Application Enquiry Form UoM – Pre-Application
• We strongly recommend you contact the project supervisor after completing the form to speak to them about your suitability for the project. You can find their details on the project descriptions –  PhD projects EPSRC Centre for Doctoral Training in Skills And Training Underpinning a Renaissance in Nuclear (SATURN).
• If your qualifications meet our standard entry requirements, the CDT Admissions Team will send your enquiry form and CV to the named project supervisor.

Our application process can also be found on our website: Apply EPSRC Centre for Doctoral Training in Skills And Training Underpinning a Renaissance in Nuclear (SATURN) If you have any questions, please contact [email protected]

Equality, diversity and inclusion

Equality, diversity and inclusion is fundamental to the success of The University of Manchester, and is at the heart of all of our activities. We know that diversity strengthens our research community, leading to enhanced research creativity, productivity and quality, and societal and economic impact.

We actively encourage applicants from diverse career paths and backgrounds and from all sections of the community, regardless of age, disability, ethnicity, gender, gender expression, sexual orientation and transgender status.

We also support applications from those returning from a career break or other roles. We consider offering flexible study arrangements (including part-time: 50%, 60% or 80%, depending on the project/funder).

 Saturn_Nuclear_CDT