Additive manufacturing processes such as laser powder bed fusion (LPBF) and Electron Beam Melting (EBM) deal with large scales of data. Components can have a build envelope of 30 cm, the material melt pool is only around 100 microns, while the laser needs to travel 10s of kilometres on its path to build a full part. Combined with cooling rates of 1 million degrees per second, a vast array of variables need to run seamlessly, quickly and leave little margin for error to create accurate and predictable components.
By their nature, additive processes like LPBF can create structures with more complex geometries, manipulating material properties to build them point by point. This means localised areas can vary within a single part. For example, the hardness or stiffness can vary throughout the microstructure, or pores can be included in certain regions of the part. Tailoring properties location by location in any given part can optimise the efficiency and performance of the components, but can be difficult to achieve – and produce consistently.
Developing a simulation process that can accurately predict the outcomes of the LPBF process will help to drive the adoption of the technology across regulated industries like aerospace and automotive when it comes to producing safe and accurate end-use parts consistently. For these simulations to be viable, the results need to justify the added cost. It needs to be accurate, timely and able to predict the performance of components to be worth the added investment. The best simulation in the world is of no use if it takes months to complete versus the 2-3 days it takes to print a prototype that can be tested.
Multi-scale simulations
We have learned through research that to predict the performance of a component accurately, specific features need to be considered. For example, if a small defect like surface roughness in a particular location of a part is missed out or ignored during the simulation process, it can alter the overall performance at the component-level completely. Simulations, like the processes themselves, need to be multi-scale to provide accurate results. For data-heavy processes like LPBF and EBM, however, the challenge is to find and maintain the balance between accuracy and computation time, so a simulation remains viable.
This balance is very difficult to achieve, especially when it comes to addressing the smaller scales of a single melt pool and the minute details contained in the microstructure, but we are starting to see some promising results. For example, we can now calculate how well a component will perform based on its material properties by deriving these from sample parts.
The ultimate goal is to find a way to connect all of the different scales within a component’s design – including the part-level, multilayer and melt pool – into a complete framework, which is not as simple as it may sound. By achieving that all important balance and tackling the issue of multi-scale data processing, simulation technology will be mature enough to create value out of the LPBF and EBM processes.
Future vision: real time simulation
Simulation is normally viewed as part of the design process; as something that is required before the manufacturing takes place. It is seen as a tool to verify that the part set up to print will come out well and perform as expected. This perception is changing, however. Future steps will include using simulation tools to support the control of the printer itself.
By moving the interaction between the simulation and control software to the machine level, it becomes part of the manufacturing process, helping to ensure that the component is produced with the expected conditions every time. This shift lays the groundwork for the ability to perform simulations in real time in future – essentially testing the components during printing to allow for modifications and corrections to be made as part of the process.
Real-time simulation is still a technology of the future, as calculations need to be fast enough to correct the local process parameters during the additive process. We are limited by hardware currently, as computers aren’t fast or powerful enough yet, but we can simplify the simulations to tackle the issue of speed in the meantime. Using modelling to adopt a semi-analytical approach allows us to speed up the simulations significantly, making them more viable and of interest to a range of regulated industries including energy, aerospace and, perhaps most surprisingly, automotive.
While speedy, accurate simulations will not necessarily shift AM from being primarily a prototyping technology to a production one for the mainstream automotive industry, there is growing interest in the ability to test functional prototypes as final components to gauge performance level. Using a reliable, multi-scale simulation process to achieve an accurate prediction of how well end-use parts will perform, can therefore help to fuel further opportunities for prototypes and the adoption of additive processes in the automotive and other mass producing industries.
Michele Garibaldi - Senior Research and Technology Development Engineer at Siemens Industry Software
Michele is just one of the expert speakers us this year’s Additive International summit, which runs from 10th-11th July at the Nottingham Belfry.
Red Bull makes hydrogen fuel cell play with AVL
Formula 1 is an anachronistic anomaly where its only cutting edge is in engine development. The rules prohibit any real innovation and there would be...