PrintSyst announced they’ve successfully integrated their AI engine into DNA.AM’s manufacturing execution system.
“MES” is a term becoming more familiar to those in the additive manufacturing world, as these complex software systems attempt to organize the complex data flows and control systems at a manufacturing site. UK-based DNA.AM offers a sophisticated MES that is tuned for additive operations. They say:
“DNA.am manages AM build planning and provides complete powder traceability, with batch splitting, control of blending and powder recovery management.”
One of the critical steps in additive production is the specification of the parameters for each 3D print job. Normally this can be quite challenging, as there are many things that can go wrong, particularly with metal 3D printing processes. Typically additive manufacturing sites will employ a squad of engineers who can use their expertise to tune these parameters carefully to obtain optimal results.
That’s usually an iterative process as, for example, heat flows will vary considerably depending on the geometry of the 3D model in question. Support structures will be tweaked to ensure proper flow of heat and tools will be used to understand how the print will warp when the supports are cut off after printing.
This part of the manufacturing process can be extremely expensive, and it’s not just due to the cost of materials and print run times. It also costs for the labor required of highly-paid engineers to perform complex job setup.
This is where PrintSyst comes in. The Israeli company had an intriguing idea to apply AI techniques to this problem. They correctly believed that it would be possible for an AI engine to “recognize” a given 3D model and suggest optimal print parameters.
Then they built it.
The result, called the 3DP AI-Perfecter, is actually able to do this. Let me explain how this works.
You may be familiar with image recognition systems, which, for example, can identify what’s in a picture. That’s a Lion, or a Pepperoni Pizza picture. Basically these AI systems have been trained through thousands of examples to categorize the digital image pattern into “Lion” or “Pizza”.
The magic thing here is the image data is simply a stream of bits, and so these techniques can apply to ANY stream of bits, image or not.
PrintSyst then arranged to present their AI system with digital input representing the 3D model, materials, input print parameters and quality of the result. The system would then train on these combinations to eventually become expert at “recognizing” the optimal print parameters.
Then it was straightforward to reorganize that knowledge to enable the near-instant presentation of optimal print parameters for a given 3D model and material.
The idea here is to simplify the work of those expensive engineers by giving them a highly optimal starting point for the print job that would greatly lessen the need for expensive iterative experimentation to dial in print parameters. While PrintSyst says the system:
“suggests printing parameters that will have the highest probability of right-first-time Additive Manufacturing builds, accurately estimates 3D parts costs, recommends on the most suitable materials to be used based on 3D parts’ functional needs and eliminates the need for trial and error.”
It’s likely that workshops will still iterate even with this tool, but just do so in far, far lower counts. That could save a considerable amount of money for each project, and thus make the operation more efficient.
About a year ago PrintSyst announced they were working with DNA.AM to integrate their AI system into DNA.ME’s MES. Now it seems they have completed that integration and the new MES becomes far more powerful. PrintSyst says:
“The two companies have partnered to develop an integrated MES that will leverage PrintSyst’s world class AI engine, which enables an automated pre-printing workflow and thus assists customers in industries such as Aerospace, Automotive and Defense, to significantly improve their productivity and scale up their 3D printing production.”
With the large number of industrial companies now entering the additive space for the first time, this could not come at a better time.
