Robotics for nuclear manufacturing1 September 2015
Dave Stoddart showcases some of the work being carried out at the UK’s Nuclear Advanced Manufacturing Research Centre to develop and demonstrate the potential of robotic systems for nuclear manufacture.
From automotive to pharmaceuticals, robotic manufacturing can save time and cost while maintaining a high and consistent quality of manufacture.
Manufacturing for the civil nuclear sector is, however, dominated by complex one-off components and small batch production, so it is hard to justify the complex programming requirements and lengthy set-up times required by robot systems.
But recent advances make robotic systems a lot more flexible and economic, and they are making inroads into the nuclear manufacturing sector.
Here at the Nuclear Advanced Manufacturing Research Centre (Nuclear AMRC), we are working with technology developers and manufacturers along the supply chain to develop and prove the value of robotic technologies. The results should help suppliers reduce cost, improve quality, reduce lead- and cycle-time, and reduce risk in nuclear manufacturing.
Nuclear reactors require some very large steel components, made with high precision and material quality. They have to be made using very large machine tools, which are expensive, occupy large swathes of shopfloor, and limit the ability of the factory to alter production.
One solution, already deployed in the oil and gas sectors, is to replace the big machining centre with portable robots. But these robot platforms are less rigid and less accurate than standard machine tools - by several orders of magnitude. And because they are used in an open environment, they are unable to deploy the flood of coolant required for high-speed machining.
At Nuclear AMRC, we are developing techniques for robotic machining using a Gamfior milling spindle mounted on a Fanuc F200i hexapod platform. To overcome the problem of rigidity, we performed a detailed dynamic analysis of the robot throughout its working envelope and under varying machining parameters. We developed a model that allows us to choose machining parameters at different locations within the working envelope.This helps create stable cutting conditions, minimising vibration and chatter.
To improve accuracy, we are trialling various metrology technologies. The robot is programmed to move through the toolpath required to produce the component geometry. The metrology system then measures the actual position of the robot in real time, and compares it to where it should be. The difference is used to correct the robot toolpath and get much closer to the nominal component geometry. Initial tests have shown that robot accuracy can be improved from hundreds of microns to tens of microns.
To remove heat from the cutting edge without covering the factory floor in hazardous liquid coolant, we are trialling minimum quantity lubrication (MQL). This is a common approach in other industries. The MQL is delivered between the cutting tool and component as very fine mist, which largely evaporates as it removes heat from the cutting process, leaving an easily manageable film of oil. We are working with coolant specialist Houghton, a member of the Nuclear AMRC, to investigate various MQL options with a view to maximising tool life.
The next stage will be to investigate cryogenic coolant techniques, where nitrogen or carbon dioxide gases are deployed through the cutting tool to remove heat. This could massively increase cooling rates and allows much faster metal removal, minimising residual stresses caused by the machining process.
The robotic machining research is funded by the EPSRC as part of the NNUMAN programme, delivered in partnership with the University of Manchester.
Cladding large components with speciality alloys helps maintain corrosion resistance and surface quality over a long service life in harsh conditions. Arc welding processes are typically used to clad the base material with speciality stainless steels, nickel-based alloys or more exotic materials. However, arc welding creates a large heat-affected zone, which can alter material properties and cause distortions in the component.
Using laser cladding instead of welding can greatly reduce the heat input. Metal powder is blown into the path of a laser; the powder melts in-flight, then solidifies on the substrate material. By altering the focus of the laser beam and re-melting the clad surface, in many applications you can produce a smooth finish that does not need machining.
The Nuclear AMRC is working to prove laser cladding is viable for civil nuclear applications, using a 15kW diode laser on a gantry-mounted robot arm. The robotic control precisely controls where the cladding material is deposited, and has enough flexibility to clad complex geometries.
Our experimental programme aims to optimise laser cladding techniques on a wide range of nuclear component materials.
Bulk additive manufacturing
Nuclear AMRC is also exploiting the flexibility of autonomous welding robots to develop additive manufacturing techniques, aiming to build near-net shape parts and features from the ground up.
We are testing a 10x5m additive manufacturing cell with a gantry-mounted six-axis Kuka robot arm and a two-axis manipulator. Working directly from a CAD model, the robot will be able to lay down weld material to form three-dimensional geometries. The cell deploys a 'toptig' welding system, which integrates the wire feed into the welding torch, developed by Air Liquide specifically for robotic welding applications.
We are developing this technology to add nozzles and features onto large forged vessels. The forging can be made smaller and simpler, saving materials and machining time.
Robotic systems can also make moving large heavy components around the factory a less complicated, time-consuming and hazardous operation.
Nuclear AMRC's sister centre, the AMRC with Boeing, is developing 'Factory 2050', the UK's first fully reconfigurable assembly and component manufacturing facility for collaborative research. One of its research topics is the use of automated vehicles to move components.
The vehicles will have an on-board autonomous safety system, so shopfloor staff can safely work around moving robot carriers. Eventually, the vehicles will be linked to the production management system so that they will automatically know when and where to collect and deliver components.
Large nuclear components also present challenges in final part inspection. Traditional co-ordinate measuring machines (CMMs) can be expensive, unwieldy and slow. Automated inspection techniques deployed by robots can save time, although achieving the same precision is a challenge.
One technique being investigated by the Nuclear AMRC is photogrammetry. A robot arm carrying a number of cameras captures images from all angles around the component. Software then correlates the images and builds a detailed three-dimensional model. This technique is not as accurate as traditional CMM inspection. But it can be orders of magnitude quicker and in most cases at adequate accuracy.
Currently available photogrammetry systems are well proven in other industries, and we are investigating for the best way to program and deploy them for nuclear components. Commercial systems do, however, come with a hefty price tag. To open up the techniques for the supply chain, we are investigating using cheaper consumer digital cameras with bespoke coded software.
The next phase of this research will be to prove that these techniques are suitable for harsh environments, so that they can be used to support the decommissioning sector in applications such as automated through-life monitoring of waste containers.
Integrated manufacturing cell
As a longer-term research goal, we want to integrate these techniques in a single robotic manufacturing cell.
Starting with a 3D CAD model, a robot arm carrying a toptig welding system will create a near-net shape part using bulk additive manufacturing techniques. The robot will then switch its welding end-effector for a photogrammetry head, and carry out a detailed inspection of the workpiece.
The cell's software will compare the scanned information to the nominal CAD model. Calculating the differences, the system will use CAM software to develop the machining tool paths needed to trim the component to its final dimensions.
The robot will then use a milling spindle and selected cutting tools for automated finish machining, before swapping back to the photogrammetry head for final part inspection.
This might seem revolutionary for the nuclear manufacturing industry, but all of the element technologies and techniques are in development.
About the author
Dave Stoddart is robotics project manager at the Nuclear Advanced Manufacturing Research Centre (namrc.co.uk), part of the UK's High Value Manufacturing Catapult. He previously worked for Sellafield Ltd.