Automotive, Rail & Marine: Steel the show

Computer modelling techniques have been put to good use in an effort to reduce the distortion of metal for ships

category/automotive, rail and marine
winner/ironing out the bumps/bae systems surface ships/
strathclyde university/malta university/newcastle-upon-tyne university

Ships have been made from welded steel for decades now, but the technique has its drawbacks. Whenever metal is heated up, it softens and can lose its shape. For shipbuilders, distortion on welding is a major problem, especially when using a thinner steel plate less than 8mm thick; any parts that have distorted after welding need to be reworked to return them to their correct shape, which adds time to the project and increases the costs.

A team at BAE Systems Surface Ships – starting in its previous, pre-acquisition incarnation as BVT Surface Fleet – decided to apply computer modelling techniques to this problem to see if there was any way to reduce the distortion. ‘The thinner the sheet, the more susceptible it is to distortion,’ said Norrie McPherson, project team leader. ‘When we started the project five years ago, in the specific project we were working on at the time, which was building Type 45 destroyers, 42 per cent of the ships were made from thin sheet, so we were looking at a very large amount of potential rework.’

The shipbuilder, based in Glasgow’s shipyards, contacted computer simulation specialists at the local Strathclyde University and at Newcastle University to look into the problem. Strathclyde has expertise in finite-element modelling (FEM) and, with a £440,000 grant from the Engineering and Physical Sciences Research Council, Prof Tom Gray’s team set about using this technique to model and predict how thermal and mechanical effects could warp metal as it was being welded.

Newcastle, meanwhile, applied another technique: artificial neural networks (ANNs). Also with additional government funding – in this case a £130,000 grant from the Department of Trade and Industry, now DBIS – a team under Prof George Bruce started to build up a system that could learn from experience how distortion could come about.

The Strathclyde team was breaking new ground, because thermal and mechanical FEM had not been used together to this scale before, said McPherson. The system that the team developed was adapted significantly from existing models and the results validated using thermal imaging on sheets as they were being welded. ‘You can have thermal and mechanical conditions happening at the same time, so by combining the two we found that we could predict a level of distortion that you would get under certain given conditions.’

This was an indicative result rather than an absolute one, he added. The process of rolling steel into sheets sets up stresses in the steel that cannot be detected or measured, but these residual stresses can cause the sheets to distort on welding. This cannot be predicted, according to McPherson. ‘But as an indicative tool, the model has been very useful,’ he said. ‘We had a fairly high level of confidence in what it was predicting or indicating, because we’d done so much validation of the model.’

The ANN was a complementary tool to FEM. ‘You put a lot of information into the model and it goes through a lot of what-if iterations, again indicating the level of distortion under certain conditions,’ added McPherson. To train the model, the team would do 60 to 70 test welds, changing the conditions slightly each time, and feed the results into the ANN.

“The thinner the sheet of steel, the more susceptible it is to distortion”

The results from using both systems gave the welding team an indication on how to avoid distortion. This, McPherson explained, is all about managing the heat of the welding process. ‘You need to put it in the right amount of heat in a controlled manner, spread it across the sheet and apply it in the right places and, in that way, you minimise the distortion,’ he said. ‘We aren’t breaking new ground in welding techniques and we aren’t having to invest in new equipment; we’re understanding the factors that contribute to distortion better and stretching the capabilities of the existing welding technologies. We’re also seeing benefits through the reduction of welding consumables. You manage the process more effectively and it gets progressively better as you feed the simulations more data. These are learning systems.’

One of the unexpected benefits is that specific factors made steel more vulnerable to distortion. ‘We knew some of these already, such as the thickness of the sheet and the heat input, but some of the chemical composition aspects of the steel showed up that we had no idea about at all,’ added McPherson. The system showed that the carbon content of the steel had a very strong influence on the propensity of the steel to distort – the more carbon present, the less likely distortion was to occur.

‘We managed to tie that up to known metallurgical aspects of the steel; we could relate it to ductility, which, in turn, was related to distortion sensitivity,’ he said. ‘Carbon tends to make steel stiffer. It should have been obvious that it would have an influence on welding distortion, but it was so obvious that nobody had realised – as far as I know, it isn’t stated anywhere in the literature.’

McPherson’s team implemented its findings throughout the Type 45 build programmes and found that it was paying dividends. Using standard welding techniques, the team could have expected rework to use up 25 man-hours per tonne of steel, which would equate to 59,000 man-hours on a Type 45. Implementing its findings from the project, 15,000 man-hours were spent; on the next ship, this came down to 13,500 man-hours. This represents a 75 per cent reduction in rework costs. ‘But unfortunately, you’re never going to eliminate this problem completely, because of residual stress,’ he said.

The team worked well together, added McPherson. ‘It was an extremely cohesive group, with much more focus than many groups I’ve worked with on such a complex project,’ he said. ‘We had some autonomy over decisions on whether we would carry out rework and that saved us a significant amount of money.’

“Combining thermal and mechanical conditions, we could predict distortion”

The team is now working on other projects using thin steel and expects to use its techniques again. ‘There’s less thin steel in the new aircraft carriers, but there’s still a significant amount of 6mm sheet,’ added McPherson. ‘And after that, we’re expecting to see the use of lighter-weight materials in subsequent naval projects.’

Runners up

Civil engineering
The other shortlisted candidates in this category were:

Ricardo UK, Denso UK, Jaguar Land Rover, Brighton University
Highly commended

Two-stroke engines have a very good power-to-weight ratio but tend to be noisy and smoky; four-stroke engines are more refined and environmentally friendly but have lower specific power output. This project aimed to combine the benefits of both by designing an engine that can switch between the two modes of operation by varying the operation of flexibly driven, independently controlled poppet valves.

Delta Motorsport, Oxford University
Special commendation for environmental achievement