A Novel Approach for Robust Design Optimization & Validation in AM and Beyond
The Möbius Design Workflow
Innovative design requires innovative tools to help engineer advanced solutions. This post by Armin Lohwasser of Rafinex describes their novel optimization approach with the Möbius Design Workflow.
We would probably not disagree that Additive Manufacturing (AM) bears great potential to industrialize novel products of superior functionality without the conventional limitations from classical manufacturing. Once we realize that up to 80% of a product’s lifecycle cost*, as well as the environmental footprint impact** are locked in during the early design phases, the impact that AM bears for profitability and sustainability alike, are tremendous.
Nevertheless, the designs for the various printing methodologies often need to comply with multiple constraints such as overhangs and printing directions, or support structures coming with each specific AM process.
Whilst DfAM - Design for Additive Manufacturing - is a methodology to prepare and adapt designs to the various AM production processes, Robust Design Optimization from Rafinex is aiming at more fundamental factors of the design. Namely the design’s intrinsic engineering property to be resilient against variabilities in loads, constraints and material properties.
Think of it as if Robust Design effectively provides a ‘blueprint’ for various manufacturing strategies by making the structural properties of a design less demanding (or vulnerable) for any specific way of production. Only when design engineering goes beyond idealized conditions, can it account for real-life variability by using Uncertainty Quantification methods. The resulting product designs will then be physically robust showing a minimized response variation versus uncertain, uncontrollable factors when used in real-life applications.
Möbius - from Problem to Optimization to Validation
‘Möbius’ is a new end-to-end CAE engineering solution based on stochastic optimization technology, built by engineers for engineers to achieve optimal and reliable, yet light and validated designs for any purpose. The intuitive (cloud based) User Interface allows for design iterations to become a matter of minutes between engineering and design, with automated workflows to evaluate the optimized parts. Möbius is delivered as SaaS with efficient computation on state-of-the-art Intel Xeon and AMD EPYC processors, which leverage Rafinex’ latest adaptive local mesh refinement technology and extremely efficient optimization algorithms to get smooth and highfidelity designs at high speed, which are then ready-to-print.
Starting your journey on Möbius is easy: get yourself a browser, access Rafinex Möbius on the cloud and get going: Buckle up!
A secure login brings to your device the database of your team’s previous projects and ongoing analysis work. Let’s start something new. We import a design space (CAD assembly) and the initial concept stage design from engineering as to the design-in-progress and the assembly to be analyzed.
A CAD model in Step Format (AP214) is imported to provide the basis for the optimization set-up. Colored surfaces in the CAD are translated as surface sets for your convenience. The figure below shows a CAD assembly for an Electronics Control Unit (ECU) in order to optimize the bracket. In gray color is the fuse box and in black is the ECU body. These two volumes represent stiffness parts in the assembly. The body in blue color is the Design Space (DS) to be optimized.
The workflow in Möbius is simple and straightforward: import your geometry (CAD), followed by a set of simple, vertically arranged panels in a top-down manner for an easy and intuitive optimization analysis set-up.
Taking full advantage of the cloud, Möbius performs all tasks asynchronously in the background, meaning that you can proceed with your work in the UI while meshing and/or solving a preliminary set-up check FEA analysis. After you have imported the geometry and created the necessary surface sets for loads and constraints, you can generate a solid tet mesh as shown in figure below:
Next, we apply boundary conditions as loads and constraints, followed by the generation of loadcases. Should you want to prepare the design’s resilience for uncertain reallife effects, you may apply stochastic distribution of the load entries. Finally, we input and assign the material properties to the separate parts in the assembly. The following task is the optimization set-up. We first select the objective, which for us in this case would be to ‘Maximize stiffness’, and the volume target as a percentage of the design space.
In the optimization constraints tab, we select for various manufacturing and analysis approaches such as: casting, symmetry, keep thickness on a surface or member size and we assign the design space parts to be optimized. Oftentimes, the combination of various basic constraints e. g. casting + member size + keep thickness will help to analyze for other manufacturing methods such as: forging or plastic injection molding.
The optimization tab allows for the selection of loadcases to be considered in the optimization. It also offers simple controls of Möbius’ unique adaptive re-meshing technique, which is a very powerful and versatile feature. In this tab we can then submit our topology optimization analysis.
Your models live in the secure cloud. Be certain and feel safe, all the AWS Data resides on AWS S3, with encryption at rest and in transit, to and from your browser. The multi-tenant access is separated at database and folder levels, and adheres to our ISO27001 certification. The Möbius solver will take care of optimizing your design on a state-of-the-art computing AWS hardware. Möbius will leverage 4-8 cores to apply sophisticated local mesh refinements to achieve a robust design through a most efficient, gradient based, solving process.
You can follow the solving process with the “job manager” tool. Once the TO and shape extractions show green, the optimization analysis is complete.
Solving time usually takes from a few minutes to a few hours, depending on the number of load cases, stochastic distributions and size of the input mesh.
Results are prepared and made accessible in the Results panel to be displayed in the UI and as a download: NASTRAN solid mesh and STL file of the shape.
In order to assess the performance of your optimized geometries, Möbius automatically sets up an FEA assembly model in meshed high-order Tet10 tets for full FEA validation.
Making fast and informed decisions is critical for you as a professional engineer, so do not judge design variants by their looks, but look at their quantified FEA structural performance results.
You can compare relevant designs - submitted sequentially or in parallel for optimization - including all the work of your team in the past. This allows making quantified decisions across multiple metrics of the designs: total strain energy, stiffness, deflection or peak stress levels across all load cases.
This effective and automated workflow from optimization to structural analysis of the shapes helps to speed up the decision making with a design without the need for export/import tasks to check shapes in 3rd party FEA. Once a final design direction has been selected within the Möbius environment, users can download the optimized shape in NASTRAN format for further detailed FEA validation to include complex boundary conditions and/or non-linearities.
Go-to-Production - Ready to Print
Normally from here you’d export and reload your optimized shape into your respective CAD system.
We promise that the resulting topologies from Möbius will not require any reinterpretation or manual intervention on your side to derive an intelligible and practical form.
The process for re-CAD’ing should be straight forward along the surface and curvatures provided by Möbius. This is as close as you can get for an optimized design to be ready-to-print.
And that’s it, feel free to contact us at www.rafinex.com and have a free tour of Möbius capabilities and our service and consulting. Request a Möbius Trial: https://www.rafinex.com/try-mobius
DfAM is a reader-supported publication. To receive new posts and support independent articles, become a free or paid subscriber.