In the second in our series of blogs about deploying Polygonica in the cloud, we caught up with Elad Schiller co-founder and CTO of CASTOR Technologies Ltd, an innovative decision support system for Additive Manufacturing. MachineWorks’ Richard Baxter asked him about how CASTOR uses Polygonica’s geometry engine for high throughput analysis of large assemblies on the cloud.
What does CASTOR do?
CASTOR is a decision support system for manufacturers. Put simply it takes an assembly of parts and rapidly identifies those where Additive Manufacturing is likely to be more effective than traditional manufacturing methods.
Our aim is to help manufacturers reduce costs by utilizing the benefits of 3D Printing. As we’ve talked to more and more manufacturers it’s become clearer that many just don’t even know where to start. CASTOR can rapidly give them a pretty in-depth answer to that question.
We are working with the automotive supply chain, manufacturers of heavy machinery and others such as Schneider Electric and Stanley. We also have a white-label version customized for printer manufacturers such as Nexa3D and materials suppliers such as Evonik.
Decision support is something new to me. From a technology perspective can you tell us some of the innovations within CASTOR’s product?
CASTOR is cloud-based. It doesn’t just assess individual parts but entire assemblies, or even entire product catalogues, in one go. A user can upload a CAD assembly and a corresponding list of materials taken from the CAD system and CASTOR will analyse all the individual parts and also look at adjacency to find where it’s effective to merge multiple parts into a single printed part.
To do this we’ve had to invest a lot of effort into building a scalable and robust back end that allows our customers to screen thousands of parts in one go. I’m really proud of what we’ve already achieved here.
So how does Polygonica fit in?
We licensed Polygonica alongside HOOPS Exchange from Tech Soft 3D. Exchange gives us fast access to pretty much every major CAD assembly file format available along with important design information such as intelligent PMI and GD&T.
We use Polygonica as one of our main geometry crunching engines. The quality of Polygonica’s automatic STL healing seems pretty widely known in the AM community but we have to go a lot further, investigating many aspects of printability such as layout, lattice and support creation. As well as healing and analysis we use Polygonica’s decimation, hollowing, offsetting, Booleans and geometry creation routines such as extrusions and sweeps.
Our Enterprise product provides weight reduction analysis and also “likelyhood of failure” checking based on Finite Element simulations. Polygonica’s mesh fixing is a key part of those workflows.
How about using Polygonica in the cloud?
Sure. Well Polygonica is C … and HOOPS Exchange is C++ … so we had to put some effort into containerizing both libraries. Of course Tech Soft 3D have gone a lot further in making Exchange friendly for cloud deployment but we didn’t find it difficult to get Polygonica working.
As you know we did encounter some rendering issues when we were first using Polygonica’s internal graphics in our early prototypes - I think you said the problems were because it was running on a VM in the cloud with no GPU attached. But we understand that’s a rare scenario for you and we appreciate the Polygonica team working hard to turn around that fix for us.
Polygonica’s internal file format (.pgs) is also pretty useful for passing data around in a cloud architecture – it’s fast, it keeps things in double precision and it also automatically preserves all our Polygonica state – most importantly the unique identifiers we attach to the Polygonica faces so we can keep the Polygonica model synchronized with the original input model.
Of course, it’s native to Polygonica, not an open format, which is a pain. If you want to use it to pass data to a part of your process that wouldn’t otherwise use Polygonica you’d still have to embed the Polygonica libraries into the consuming process.
Are you planning to look into any other Polygonica features?
Well, the feature detection looks interesting – if the original CAD part isn’t available then detecting holes and pockets can give us some useful extra information. Also the tunnel identification and extraction feature looks like it might help with some of our topology analyses.
What about your experiences with Tech Soft 3D, MachineWorks Ltd and software components in general?
Heh heh. Well, of course, we have to use components. We’ve found using commercial components can really accelerate things as long as the cost can be justified.
In most cases we’ve been really impressed by the quality of the engineering and testing that goes into each release of Exchange and Polygonica. Nothing’s perfect but I don’t really expect it to be. Exchange supports a vast array of formats and Polygonica offers a vast array of geometry operations. What’s exceeded my expectations has been the responsiveness of the support teams and the fact that our inputs, feedback and requirements are clearly valued.
What next for CASTOR?
Well. Specifically we are going to take a deep dive into metal printing to help manufacturers really understand the opportunities - and limitations - they will face. We’re also hard at work integrating CASTOR into customer workflows to provide a more seamless automatic and faster experience.
CASTOR is bringing together best in class geometry and simulation tools with machine learning and applying them to really interesting areas of Additive Manufacturing in ways we don’t think anyone has tried before. Keep watching!