Additive manufacturing, commonly known as 3D printing, is widely used for the manufacture of plastic prototypes and consumer products. However, the various additive manufacturing (AM) processes available are still in their infancy and face challenges to enable the mass production of parts that can pass performance guarantees required by industry.
Simulation provides key insights into design and build decisions to push the different AM processes out of the research arena and into mainstream industrial production, but process reliability and repeatability must be ensured to make this a reality, as Alan Prior, vice president SIMULIA Worldwide Center of Simulation Excellence at Dassault Systèmes, said: ‘Physics-based simulation offers a way virtually to engineer the process to minimise unwanted effects like stress and distortion in printed parts, so that successful prints can be achieved without time-consuming and costly test iteration in the real world. This will exponentially accelerate progress towards widespread adoption.’
One of the core challenges for AM process simulation is the multi-scale nature of the problem. Alonso Peralta, principal investigator at technology house Honeywell, explained: ‘The difficulty that arises with numerical simulations in additive manufacturing is the large temporal and spatial scales that are transcended in the problem.’
There are many different AM processes that vary in their method of layer manufacturing. Powder bed fusion methods use either a laser or electron beam to melt and fuse material powder together. Dr Peralta said: ‘In this process, we are trying to melt powder particles that are less than 100 microns in diameter, and we have to build parts that are less than a metre in size. Furthermore, we must melt the powder where the interaction time of the laser and the powder is in the order of a few microseconds and it takes up to a few days to make a complete part.’
This increases the complexity of the simulations, as Mustafa Megahed, manager of CFD and Multiphysics Centre of Excellence at the ESI Group, explained: ‘Models also have to deal with multiple physical aspects, such as heat transfer and phase changes, as well as the evolution of the material properties and residual stresses throughout the build time. The modelling task is, therefore, a multi-scale, multi-physics endeavour calling for a complex interaction of multiple algorithms.’
Another limitation is the lack of well-characterised material data to understand detailed phase transformation effects and fatigue responses. Experimental data is aiding development in this area, alongside molecular-level material simulation, according to Prior: ‘We can create atomistic models of material concepts to study these effects, and through our continued development efforts we will link this molecular scale simulation with part-level simulation.’
Computational consulting company AltaSim Technologies is focused on the laser powder bed fusion of metallic powders to produce metallic components for aerospace. This comprises two research areas: shape optimisation, which aims to produce a component at a minimum weight while meeting target mechanical properties, and AM process physics, with studies ranging from fundamental evaluations of the laser beam-particle interaction to estimates of residual stress/distortion in AM components.
Simulating AM process physics is a challenge, says Jeffrey Crompton, principal at AltaSim Technologies: ‘Of these, shape optimisation is perhaps more readily handled with existing capability whereas analysis of AM process technology is complicated because of the wide spread of deposition schemes available and the complexity of the interdependent phenomena that need to be included in any analysis.’
The range of issues associated with AM process physics is extensive, from determining the best powder size distribution to minimising occurrences such as porosity and key-holing during laser scanning to cracking and microstructural control during component fabrication. This range of issues affects the simulations, as Crompton said: ‘It depends what level of detail is require to look at the phenomena of interest. The more detail that is required, the longer the analysis times and at some point it becomes quicker to run the test than do the analysis.’
‘The Holy Grail is to produce thermal histories that can create the required microstructure/properties and overall structure. At the moment simulation is not at a point where it can do this.’
Additive application areas
AM is popular in the aerospace and biomedical sectors, with the automotive and the sporting goods industries also showing interest. Weight reduction, part customisation and structural integrity are the main drivers for these sectors, but quality is also a key concern, as Mustafa Megahed, manager of CFD and Multiphysics Centre of Excellence at the ESI Group, said: ‘If you are producing a component that is going to be placed in an aircraft, for example, you have to be able to guarantee the quality of that component. These industries need performance guarantees.’
Topology optimisation is a mathematical approach used to simulate AM processes and meet such stringent requirements. It optimises the material layout within a given design space, for a specific set of boundary conditions and loads, so that the resulting layout meets predefined performance targets. This allows engineers to find the best concept design to meet these prescribed requirements. It also replaces costly and time-consuming design iterations.
Research at the University of Pittsburgh focuses on the development of efficient topology optimisation algorithms for lightweight design of additive manufactured components by incorporating cellular structures. Albert To, associate professor at the University of Pittsburgh, said: ‘The main challenge to the industry is exploiting the enormous design space provided by 3D printing and AM while addressing various manufacturability requirements. My group and others have developed numerical algorithms and software to tackle this challenge, but we are still not there yet.’
Robert Yancey, vice president of additive manufacturing at Altair, said: ‘Without topology optimisation, 3D printing is just a new manufacturing method. With topology optimisation, 3D printing is a technology that allows us to design and build structures that are similar to what we see in nature – elegant, structurally efficient, aesthetically pleasing, and tuned to meet precisely the intended purpose in the environment where it exists. That is very exciting.’
Altair’s topology optimisation software, OptiStruct, helps engineers take full advantage of 3D printing by generating elegant organic designs that are structurally efficient and lightweight. Yancey said: ‘OptiStruct coupled with metal AM methods is showing weight reductions as high as 80 per cent over traditionally machined parts while, at the same time, guaranteeing that components withstand the required loads.’
Another benefit of AM is to reduce the number of single parts in a system, leading to further weight and cost reductions. For example, the biomedical sector could use AM to create individually customised products such as prosthetic limbs or surgical aids. That’s not all, as Yancey added: ‘Customisation can also come into play when looking at individual solutions for bikes or other sporting goods.’
On your (robot) bike
The Robot Bike Co. R160 mountain bike frame, unveiled earlier this year, allows each frame to be tailored to a customer’s individual measurements or specifications.
Altair’s suite of simulation products helped to optimise the bike’s connectors, as Ed Haythornthwaite, co-founder of Robot Bike Co, explained: ‘After developing the suspension kinematic, construction design methodology and aesthetic design, we utilised SolidThinking Inspire, in conjunction with Altair ProductDesign team, to enable the detailed design through stress-based topological optimisation. This gave us the confidence that we had a robust yet lightweight design based on the load cases acting on the bike.’
SolidThinking Inspire allowed the team to take the existing designs and apply a variety of loads to which the bike frame would be subjected during real-world rides. This data was then used to generate a geometry layout that removed material where it was not required to meet performance targets. Haythornthwaite added: ‘The team took this geometry and used OptiStruct to provide further refinement to material thicknesses. Throughout this process, the designs had to be optimised for the AM process, which included determining the ideal print angle and placement of the supporting structure to avoid the component collapsing during the manufacturing process.’
Haythornthwaite said: ‘SolidThinking Inspire allows us to benefit from the flexibility that AM offers, all without expensive tooling. It means that we can offer this unique customisable bike frame, which would have otherwise been too expensive to manufacture on an individual user basis. The software allows us to speed up the entire design verification process and gives us confidence that our products will stand the test of time.’
Honeywell is involved with several AM programs. Some programs, such as DARPA Open Manufacturing, are developing a suite of tools to simulate the process. Others are developing technologies to help understand what it takes to make specific products to meet aerospace requirements. Peralta said: ‘The overall goal is to understand the physics of the process and to define the process parameters that yield a part meeting dimensional and structural requirements to meet all aerospace safety and quality standards.’
These programs have used simulation software from the ESI Group to model the powder spreading, followed by the melting of the powder with all the different phenomena governing the fluid flow, and ending with the solidification of the molten metal. Solidification results in high residual stress caused by the large thermal gradients that develop during the process. Peralta said: ‘We have been able to simulate the melting of the powder, including all the relevant physical phenomena that play a part on the formation of the melt pool size and shape, and then we have been able to simulate the deformation that arise because of the residual stresses that develop during the process.’
The huge range of phenomena to simulate means that challenges remain to understand how most efficiently to obtain the correct solution. Peralta added: ‘The biggest challenge has been to understand all the knobs that need to be turned in the software to model the correct physics to obtain the desired results and, once we obtain those, to be able to interpret them correctly given that often, the obtained behaviours go against our preconceived notion.’
Designing the future
Engineers at the Manufacturing Technology Centre (MTC) take prototypes from the research arena and use industrial equipment and cutting-edge simulations to mature these designs. They work with a range of materials, including steels, titanium, aluminium alloys and nickel super-alloys, to create complex geometries through AM. The engineers have used simulations to optimise a particular AM technique known as shaped metal deposition (SMD).
In contrast to powder-based AM techniques, SMD has the capability to build new features on pre-existing components or allow a number of materials to be integrated into a single part. It is also a faster technique, but not as refined as powder-based processes.
Similar to fusion welding, a mass of molten metal is incrementally deposited on a surface. However, components manufactured through SMD contract as they cool down, causing internal tensions to accumulate, which results in component distortion, such as bending, when a straight edge is desired.
Two strategies can be used to overcome these distortions. One method is to predict the outcome of the design by modifying certain build parameters to minimise deformations and the other involves changing the design to counteract such deformations.
Engineers at the MTC used COMSOL Multiphysics to turn an existing model into a simulation app, which is based on a thermomechanical analysis of thermal stresses and deformation resulting from SMD thermal cycles. It can predict whether the resulting component will pass a range of acceptance criteria, based on the specific nuances of the application area.
A major benefit of the simulation app is its intuitive user interface, which means individuals and teams without a background in simulation can run their own tests and obtain results. Borja Lazaro Toralles, advanced research engineer of digital engineering at the MTC, said: ‘This meant that complex simulation models that usually required a specialist to run them now had an interface with a front-end that a designer could use. The designers can work out how to manufacture a product by editing certain parameters, which frees specialist resources and gives a wider range of alternatives to the designers.’
Within this app, users can test different geometries and materials, alter the deposition path, change the heat source or apply varied meshing sequences.
The next step is to link a range of different applications, including COMSOL Multiphysics, CAD, and statistical analysis software, into one integrated workflow with an equally intuitive user experience. Lazaro Toralles added: ‘The idea is that we would like to produce a simplified interface in the same vein as the simulation app we produced using COMSOL.’
Incorporating multiple pieces of software that are not designed to work together is more tricky than the simulation app development, but the team are working to make the workflow more robust and are working with designers to improve the interface, as well as taking a little design inspiration from the apps you find on mobile phones.
The engineers are also endeavouring to improve simulations to more accurately predict specific properties in a range of materials – for example, to predict when cracks may appear in certain structures fabricated using SMD.
There is still a long way to go to fully realise the potential that additive manufacturing could offer the industrial world. Simulations are the driving force behind the push to make AM a mainstream technology, but such computational techniques must improve to meet the massive multi-scale issues and simulate the sheer range of phenomena required to provide a coherent picture.
It’s a huge undertaking, but the potential advantages are equally awesome.
