Session: MF-06-02 Materials and Technologies for Nuclear Power Plants-2

Paper Number: 154602

154602 - Bayesian Optimisation Applied to a Thick-Section Joining Process 

Abstract: 

Joining sections of nuclear pressure vessels is performed using an established arc-based welding process. Sections of some vessel components can be in excess of 200mm and it can take many weeks, even months, to perform a single circumferential weld. With multiple joints per vessel and multiple vessels in a typical pressurised water reactor (PWR) application, the joining process is often on the critical path.

Over the last 15-20 years, there has been significant advancement in the development and deployment of electron beam welding (EBW). Already an established and proven technology (TRL6+) in aerospace, medical and automotive sectors, work to introduce the technique into the nuclear sector has been progressing steadily. The key advantages of EBW when compared to arc-based processes are numerous and significant: EBW is a single-pass, autogenous process, negating the need for inter-stage non-destructive examination (NDE), filler material, interstage cooling and re-heating, and elimination of the requirement for inert gas. Trials have also demonstrated that a single circumferential weld on a 100mm thick section can be reduced from 100 days to less than 10 days, with beam-on typically less than 1 hour.

Electron beam welding does require a vacuum to provide the focused intensity of the beam. Removing the air molecules reduces beam scatter due to electron interaction. Traditionally a vacuum chamber is used and the EBW capability and the components to be joined within this chamber. Due to the size of the components used in the nuclear industry, vacuum chambers are prohibitively expensive or complex. Recent developments in ‘local vacuum chamber’ EBW has been successfully demonstrated on a number of applications, eliminated the need for a large chamber and providing the necessary dynamic vacuum system to join thick-sections using the EBW process.

The EBW process is governed by a number of key parameters, including the beam power, transition speed of the beam and vacuum level. During a process development programme, these key parameters are tuned in order to meet certain acceptance criteria. Finding the right set of parameters is extremely difficult because the total number of possible process parameter combinations is very high, resulting in lengthy technique maturity programmes.

Bayesian Optimisation (BO) is well-suited to developing and optimising advanced manufacturing processes, as it lends itself well both to cases where the objective function is expensive to evaluate (it doesn’t require large numbers of process runs) and effectively handles multiple output constraints.

This paper will review and report on the state-of-the-art of local vacuum EB welding in industrial applications. It will also provide a review on the potential for a Bayesian Optimisation technique to tune the process parameters for EBW trials. A similar study was presented previously at PVP (PVP2023-106320) and a parallel abstract has been submitted for PVP 2025, which aims to show how Bayesian Optimisation could be deployed to another equally complex manufacturing process, and in doing so, reducing the guesswork and trial-and-error approach previously adopted, and therefore drastically reduce number of manufacturing trials required, and consequently the time and cost of a technique maturity programme.

As part of the Bayesian Optimisation process presented in this paper, an acquisition function is defined to simultaneously account for the multiple acceptance criteria, while also optimising the process speed. The method has also been developed to account for robustness criteria in a natural probabilistic manner, so as to select process parameters that reduce the risk of non-conformity. We show how this technique is used to more efficiently tune the process parameters to achieve an acceptable result, saving time and cost in the technique development programme.

Presenting Author: Steven Lawler Frazer-Nash Consultancy Ltd.

Presenting Author Biography: Steve is a chartered engineer, a fellow of the Institution of Mechanical Engineers, with 37 years' of experience in design and manufacturing spanning nuclear, aerospace and automotive. An advocate for advanced manufacturing and realisation of innovation in the nuclear sector, Steve has led and matured a number of techniques and technologies for the nuclear sector, working with many UK and US organisations, including EPRI amongst others. He is currently leading the advanced manufacturing program, "AMTecH" at Frazer-Nash Consultancy in the UK.

Authors:

Steven Lawler Frazer-Nash Consultancy Ltd.
Greg Nelson Frazer-Nash Consultancy Ltd.
Chris Punshon Cambridge Vacuum Engineering Ltd.