Advertisement

In recent years, modern Additive Manufacturing has grown exponentially in terms of what can be achieved. It is no longer a method to just produce nice-to-look-at mockups. In fact, Additive Manufacturing has opened up possibilities for product design that only a few years ago were not even contemplated. We can now use organic design almost completely without limitations; we can manufacture almost anywhere needed: under the sea, in deep space, in crisis zones; by using modern scanning technology, we can make new parts for any situation at the push of a button. It is truly an amazing prospect.

In the real world, however, there is a manufacturing process involved. And with a technology as new as Additive Manufacturing, there are a lot of unknowns: part distortions, residual stress, repeatability & scatter, micro-structure evolutions, surface finishing etc. Brought to engineers are also challenges beside the opportunities. It is here that we can begin to translate real-life questions into relevant engineering tasks. Can we make the structure 50 percent lighter and 50 percent stronger at the same time? Can we print spare parts with in-service quality? How many parts can we print per hour if they need to exhibit a specific strength? Can we achieve durability? How can we design from single parts to full assembly?

Unfortunately, traditional design and simulation solutions are not enough to answer these questions, or moreover, make the solution efficient for product development. These challenges can only be addressed through multi-scale multi-physics cross-function platform solutions.

The first question a designer has is how to design the shape of the part/parts to meet the in-service loading condition, but also make it lighter at the same time, by utilizing the capability of Additive Manufacturing to build organic shapes. An engineer will probably give the answer: Topology Optimization. However, although Topology Optimization has been out for decades, it has never been customized to address Additive Manufacturing specific constraints, or been developed to enable smooth transitions from optimization results back to a new geometry. Engineers spend months manually reconstructing FEA mesh into a new organic CAD geometry.

But with platform technologies, things are much easier. An example we’ve seen with the Dassult Systems 3DExperience platform, designers can use a Functional-driven Generative Design app to perform topology optimization by applying mechanical and thermal loads, as well as specific Additive Manufacturing constraints such as overhang minimization. As a result, designers can reconstruct a smooth CAD geometry with one click of a button. Additionally, multi-part optimization is made available for assembly and system level designs. Figure 1 shows a multi-part optimization example for an automotive door hinge assembly. The design space includes five components that make up the assembly. Designers can preview the optimum assembly shapes, reconstruct the geometry and validate the in-service performance of the assembly, which enables them to implement the design and optimization of this assembly into their short product development cycle.

Figure 1: Assembly/Multi-Part Optimization of an Automotive Door Hinge Assembly

The design and optimization is only one part of an engineer’s story for Additive Manufacturing, as it’s a cross-function problem. While designers work to achieve optimum shapes to utilize the full potential of Additive Manufacturing, machine process specialists work on improving machine parameters for building better parts, which in turn affects the design. The Additive Manufacturing process is also a multi-scale multi-physics problem.

As such, it is essential to capture the melting/solidification physics at milliseconds in time and micrometers in length as well as overall printing processes of a typical production part at hours in time and millimetres in length. Therefore, using traditional simulation alone will not be enough to solve the puzzle completely. We need to address the challenges by using a multi-scale multi-physics cross-function platform to digitally connect end-to-end from the beginning – design – to the finish – production – amongst different experts in the organization including designers, machine process specialists and analysts.

Let’s look at a turbine blade use case (Figure 2). In another example in using Dassualt Systems’ 3DEXPERIENCE platform, engineers can prepare their turbine blade build process, orient the part properly on the build tray, generate support structures, nest the parts, and minimize support structure by geometric optimization tools in a virtual machine environment. The virtual build preparation can be printed by machine by exporting into machine readable file format such as .stt, or can be taken to the next virtual print step to understand the physics of the print process in order to optimize the part/assembly design and machine process parameters for best build quality.

Figure 2: Turbine Blade: Real Print vs. Numerical Model

With different multi-physics simulation methods engineers could choose to simulate the virtual print process with either thermo-mechanical or eigenstrain methods. The outcomes could result in performing a detailed thermo-mechanical analysis to predict complex physics – such as buckling and failure – or performing a direct stress analysis based on a pre-defined eigenstrain library which will eventually become a faster approach once the user builds up their library. Figure 3 shows the equivalent results from thermo-mechanical and eigenstrian simulations on the same turbine blade design. Distortion results could be directly mapped back to create new compensated designs with the Virtual Shape Compensation tool to minimize the distortion induced from the printing process. The process simulations are scalable with simple changes in time increments and mesh size, due to essential technologies such as mesh-intersection and progressive element activation.

Figure 3: Distortion prediction of turbine blade print: thermo-mechanical vs. eigenstrain pattern based.

In this case study, the turbine blade is printed with Selective Laser Melting (SLM) process using Ti64 material. However, similar process simulations can also be performed regardless of the print process (e.g., FFF, FDM, LDED, EDAM, MJFTM, SLS) or material (e.g., aluminium, steel, IN625, IN718). Voxel meshing with STL geometries, lattice optimization, support analytics, heat treatment, machining, cracking, in-service loading, durability can all be performed. Figure 4 shows an example of support strategy and analytics. The support design affects the distortion of the turbine blade.

Figure 4: Support strategy and analytics for the turbine blade: 1 mm thickness vs 2 mm thickness

To make sure this aerospace turbine blade will sustain the critical in-service loading when brought up to the air, we need to understand the as-built material behavior by understanding how the material properties change during the print process. Engineers can look into microscale morphology and track the microstructure phase transformation during the print, and connect to the final as-printed material behavior in the platform. Figure 5 shows how engineers analyze the grain morphology at different solidification rates to design better metal alloys for the turbine blade, look at material phase transformation during the build, and connect the phase content prediction to the final mechanical properties of the turbine blade.

Figure 5: Material modelling from atoms to parts

In summary, designing an Additive Manufacturing turbine blade that can be brought up into space is, in the real world, very challenging and cannot be achieved with conventional simulations. Challenges need to be addressed through end-to-end multi-scale multi-physics simulation solutions using a cross-function platform.

Advertisement
Advertisement