Process modelling improves aero engine design

Conventionally, gas turbine discs are either bolted or electron beam welded to each other to create a drum, which is then joined to one of the central shafts. However, due to the material that needs to be added to the discs in order to bolt through them, bolted joints are heavy, which compromises efficiency and costs airlines money during operation. Being a fusion welding process, the conventional alternative of electron beam welding leaves a re-cast layer at the weld line with significantly different properties to the wrought parent metal. So, to achieve a high-integrity joint with close-to-wrought properties, Rolls-Royce uses a rotary friction welding, or inertia welding, process. The process involves spinning one component, attached to a huge flywheel, and then bringing it into contact with the other component under a large axial load. Heat is generated at the contact interface due to friction and material at the interface is extruded out as flash. As the rotation slows to an eventual stop, an extremely high-integrity bond is formed without actually melting the parent material. However, predicting the distortion, residual stress/strain state and material properties that result from this process combined with numerous heat treatments and machining operations is a significant challenge. "The whole manufacturing route has an impact on the properties of the final part," says Paul Brown, team leader for process modelling at Rolls-Royce. "For example, the residual stresses that cause machining distortion come from the quench after the heat treatment. Some of this gets relieved when you start removing material by machining, and it causes the disc to distort. We need to be able to predict this distortion so that we can design a machining process accordingly." Optimising the manufacturing process using physical tests alone is extremely expensive and time-consuming, and often does not yield the required information due to the difficulty of making the appropriate measurements under extreme process conditions. "It is, for example, very difficult to measure residual stresses in a real component, and impossible under most realistic thermal and mechanical loading conditions," says Brown. "Provided you do an appropriate amount of validation and you are confident it is predicting the right thing, you can get so much more information from a model." Then there is also the issue of what to optimise for and when in the product introduction cycle to do it. "Manufacturing isn't a constraint on design, it is an opportunity to optimise the product, process and material," says Brown. "Trying to optimise a manufacturing process once a product design is fixed is an inefficient way of going about things." Using a combination of targeted physical testing in conjunction with advanced materials and process modelling, the product design and manufacturing process can be optimised together at an early stage where the cost of design change is much lower. Rolls-Royce wants to migrate the modelling capability from the small, specialist team to the numerous manufacturing, design and materials engineers in the company so this work can be done by the people who are making critical decisions. It hopes this approach will yield more optimal processes and ultimately improve overall engine design. To achieve this, Rolls-Royce has teamed up with Wilde FEA, the University of Birmingham and the University of Nottingham and set up a project called PROMOTE (Process Modelling for Tomorrow's Engines). The project, which is part funded by the Technology Strategy Board (TSB), will further develop and then capture the fundamental modelling expertise within Rolls-Royce and the University of Nottingham and wrap this in to a user interface that is both useful to, and usable by, non-specialist engineers. "We don't want non-specialists to have to worry about a lot of the modelling specific terminology like boundary conditions, which material models to use, how to set up the meshes," says Brown. "So we try to standardise and automate as much of that as possible and develop software tools to handle complex or repetitive tasks for them." Inertia welding can be expensive and without computer modelling it takes a long time to develop components and materials suitable for the process. And aside from speed and cost considerations, models can actually tell you a great deal more information about the real manufacturing process than the physical tests. As well as designing new products, modelling can also be used to investigate potential new materials, or even process equipment. Using modelling, engineers can potentially test materials and equipment that either aren't available yet or are very expensive, prior to committing to physical prototypes. The project is structured around the use of a finite element analysis package called DEFORM distributed by Wilde FEA. DEFORM is a high-end, specialist simulation tool that was developed by US company Scientific Forming Technology Corporation. It can be used to predict the behaviour of materials that go through a chain of manufacturing processes; from forging to machining, heat treatment, quenching, joining and so on. It is then very good at predicting the residual stress and microstructure of the finished product or component. "You can use DEFORM at the front end of the design to simulate the processes that develop the microstructure," says James Farrar, business development manager at Wilde. "The vision that DEFORM has is to capture this whole process route including the product application. This project is looking at taking that to the next level of integration. Whereas an expert user in DEFORM can do this now, what this programme is looking at is delivering that capability to a non-specialist." There is, of course, a whole chain of processes where the microstructure is evolving. What the designers are interested in the microstructure at the end of the process, as that ultimately determines the mechanical properties. "It is the whole process sequence that we are interested in," says Brown. "By developing a tool for accurately predicting the as-manufactured state of the material and the as-manufactured properties of the material, we can help improve the design analysis for aero engine lifing and performance." Says Farrar: "The existing inertia welding process might be OK, but if I can optimise my heat treatment process beforehand, which has an effect on how it is welded, I can perhaps improve it still further. So it is trying to tie that whole process route together. And the technology specialists in companies like Rolls-Royce are interested in this entire process route in addition to the technology. They want to find out what is the best route to achieving design objectives whilst minimising or controlling other outcomes such as distortion." It is not just Rolls-Royce engines that could potentially benefit from this technology. "The bigger picture [within PROMOTE] is the more efficient use of aerospace engines," says David Deakin, managing director of Wilde FEA. "But the direct spin-offs are to other companies that are doing inertia welding; that could be aerospace, military, automotive, construction equipment, oil and gas, and those sorts of industries." "The other spin-off area is to anyone that wants to have non-specialists using a very advanced code in a user-friendly environment. Anyone that does heat treatment and then machining has the problem of residual stress and distortion. This project ties all these things together."