Subdivision Modeling and Topology Optimization for Automotive and Beyond
Interview with Founder and CEO of Bionic Mesh Design, Thomas Spoida
In this interview, we present Thomas Spoida, Founder and CEO of Bionic Mesh Design, a company specializing in lightweight design optimization in the automotive and aerospace sectors. Thomas shares his background, the motivations for establishing Bionic Mesh Design, and the unique approach of combining topology optimization with subdivision modeling.
Thomas discusses the challenges in integrating topology optimization results into CAD data and how subdivision modeling addressed these issues. He provides an overview of his company's typical workflow and the techniques employed at each step.
Thomas addresses the difficulties in incorporating advanced manufacturing processes into established automotive and aerospace companies by outsourcing geometry creation, and offers insights on how engineers can move from the R&D stage to production at scale.
Could you begin by sharing some information about your background and the factors that motivated you to establish Bionic Mesh Design?
During my work at a large automotive OEM I worked on many different projects related to lightweighting. For most of these projects I used topology optimization to be sure to start with the perfect shape. But the problems always started when it came to transferring the topology optimization results into ready CAD data.
I am actually a calculation engineer and at that time had no real idea what it meant to generate ready CAD data. The big disillusionment always came when I received the first CAD data from the design department for recalculation.
Unfortunately, most of them had very little to do with the organic forms of topology optimization. This was partly due to the fact that one had to make certain compromises in the shape with regard to manufacturability, but for the most part the problem lay in the design methodology itself.
Parametric CAD solid design doesn't really allow organic shapes, and if you want to approach topology optimization as closely as possible, it takes forever and the design tree in CAD becomes extremely complex. In the end, there was always very little left of the lightweight design potential and the topic of topology optimization was thus rightly called into question.
However, I did not want to be satisfied with this and looked around for alternative design options. In the process, I discovered that there was a second 3D world in addition to the CAD world I was familiar with. In animated films and computer games, there are countless 3D bodies, some of them highly complex, which also have to be created by someone and, above all, very quickly.
When I took a closer look at the methodology of the programs used for this and discovered that the same methodology, namely Subdivision Modeling (+ Subdivision Surface), is already integrated in CAD, it was clear where the journey was headed.
After I had developed several prototypes for future series components by means of topology optimization and subsequent design using Subdivision Modeling, and these were very quickly declared as series components for future vehicles, I decided to go into business for myself in this field and start my own company.
As of today, Bionic Mesh Design has 6 permanent employees plus interns.
We offer the complete process chain starting with topology optimization, design up to the support of the components at the manufacturer. But sometimes we just get topology optimization results from customers, which we re-model according to the desired manufacturing method.
Can you explain a ‘typical’ workflow, from CAD to simulation through optimization and verification with what software you use at each step and why?
In general I would like to mention that using subdivision offers a completely new way of developing components.
Everything is now based on meshes and these meshes can represent everything.
We are able to use a subdivision mesh simultaneously as simulation mesh and as CAD. That means that we can do everything in parallel. Still optimizing the final design of a component and also being able to hand over a mature design of the component as CAD file to our customers for package examination or inquiries to potential manufacturers.
Our typical workflow starts with generating a design space. Here, too, 3D modeling tools are far superior to CAD programs because you can use subdivisions to define the available space much more quickly and precisely.
After the topology optimization we use different 3D-modeling tools to create a coarse manufacturable Subd body that can be imported back into CAD to be complemented by Boolean operation for mechanical processing.
This is the point where you still need parametric design to ensure the necessary precision in drilling, for example. Since we now have the first CAD data very quickly, there is enough time to perfect the component with subsequent wall thickness and shape optimizations in terms of weight and performance.
It currently would be difficult to get this workflow that steps out of ‘traditional CAD ecosystems’ and relies on ‘sculptural’ tools into a major automotive company’s engineering process. How can automotive (and aerospace) companies innovate and adopt more advanced manufacturing processes if they cannot adopt the software that enables these new technologies?
I don't see a problem with this at all, since we already work with several large OEMs and they actually only had to adapt their workflows very little or not at all.
Very often the creation of geometries does not happen at the OEM itself and is outsourced.
The only thing the customer has to ensure is that our data can be read in as living subdivision bodies and can be further processed in CAD. This sometimes necessitates the abolition of licenses for subdivision environments in CAD programs such as NX or Catia.
On the other side of the equation, how can an engineer that understands computational design and is curious about advanced manufacturing techniques get parts past the R&D stage and into production at scale, whether they be optimized cast parts, or AM, when they cannot hand over a ‘package’
I believe that the biggest problem to date is the sequential development method and strict separation between design, simulation and production.
Real lightweight construction can only come about if all areas work together right from the start.
A small example related to cast components: You optimize the topology and look at the result together with a casting expert. This decides which optimized load paths can serve as areas for the in-gates or can already serve as flow paths for the melt. Or where perhaps these structures are missing in the optimization result. In this way, these can be defined as so-called non-design areas in a renewed optimization loop and have both structural and production-related determinations.I think that als DfAM this is extremely important.
As the transportation industry shifts towards electrification, mass becomes incredibly important as does thermal management of electronics. Topology optimization and optimized lattice structures are emerging as the best engineering solution to reduce mass in the design of components and systems, yet the cost of additive manufacturing inhibits adoption in mass production in the same way composites have stalled in the past. What can we learn from composites in automotive to not make the same mistakes in AM?
Automotive has always been very much dominated by costs. This is certainly a point where AM still has to do its homework if it wants to be considered a mass production process.
When purchasing cast components, it is often a matter of very small cent amounts that are negotiated. At AM, we are currently still talking about prices that are sometimes several hundred times higher than the costs of cast components.
The second point is speed. In mass production processes, everything is designed for very short cycle times. Personally, I cannot imagine that the current cycle times of casting and forging processes can be achieved in the foreseeable future.
Another point where composites are at a disadvantage is the issue of sustainability. In addition to the high amount of energy required to manufacture the fibers and the formation of respirable particles during processing, the issue of recycling was never solved.
When it comes to AM, I definitely see issues that need to be improved when it comes to energy expenditure.
Subscribe for irregular updates.
Let’s pretend the cost model for metal AM parts is now equal to casting, what would we need from software companies to allow automotive engineers to adopt tools optimized for the process, for existing incumbent CAD ecosystems and startups hoping to get adoption beyond R&D?
If you ask us personally, everything you need is already there. If other engineers also want to use our processes without having to work their way through manuals and tutorials for artist programs, the programs should have an engineer-understandable UI.
As we wrap up this interview, we encourage those interested in lightweight design optimization to reach out to Thomas Spoida and the Bionic Mesh Design team. Their knowledge in topology optimization and subdivision modeling can be a valuable resource for enhancing your design and engineering practices within the automotive and aerospace industries.
To learn more about the leading developments in computational, generative and AI driven design, CDFAM in NYC June 14-15 features engineers, academics and software developers at the forefront of design for advanced manufacturing.
Join us for two days of immersive presentations and networking sessions, registration is now open and space is limited.