Senvol, a New York City-based database management company that specializes in additive manufacturing (AM) software solutions, has announced that it has successfully achieved one of the key objectives of its November 2020 U.S. military contract. Army Research Laboratory (ARL) uses the company’s machine learning software Senvol ML to rapidly characterize and design additive manufacturing (AM) materials.
The National Center for Manufacturing Science (NCMS), a research and development non-profit organization, administered the contract as part of ARL’s Advanced Manufacturing, Materials and Processes (AMMP) program. Specifically, Senvol has successfully demonstrated that Senvol ML can effectively “develop statistically proven material properties similar to material tolerances.” In this case, 17-4 PH stainless steel on a powder bed melting (PBF) platform.
This means that while Senvol ML did not develop a “true tolerance” (a formally certified property such as “allowable stress” for a particular material), the software could achieve it with further development. It means that you have achieved the result you suggested. As noted by the company, achieving tolerances was primarily hampered by budgetary constraints. I’m just speculating, but it’s possible that the company showed results through simulations, and then, with more funding, they’ll be able to test the validity of the same results in wargame exercises, tech demos, etc. Maybe.
In a press release announcing the successful use of Senvol ML in material tolerance development, Senvol President Annie Wang commented: The Senvol program has been very successful in demonstrating a new approach to: [AM] Enables development using machine learning. We are very pleased with the results and look forward to continued research in this cutting-edge field. “
Lockheed Martin Fellow Hector Sandoval, who reviewed the AMMP program test results, commented: The current process works well, but has some limitations. It was very exciting to support the AMMP program by reviewing the technical approach, test results and final presentation. It was great to see firsthand the potential of leveraging machine learning-based approaches to establish material tolerances. “
In general, the pace of progress in metals qualification for AM is a very good measure of the pace of progress in this area. Related to that, a high percentage of total US military contract value related to AM this year, in addition to projects such as the Senvol ML project that are closing in and could be leveraged for larger contracts. , digitized parts inventory, etc.
First, it shows that the infrastructure is already in place to support the scaling up of contracts that have proven successful in the basic research phase. For another, with the U.S. military taking responsibility for her AM metal certification in its hands, the pace of progress in AM metal certification could follow the DoD’s Manufacturing Readiness Level (MRL) timeline. . In particular, the potential of machine learning to speed up that process gives the military a very specific strategic objective to rapidly accelerate its move to direct control of the country’s AI infrastructure. And even if other organizations argue that they have a justification for regulating the use of AI in the certification of metals used in manufacturing, the metal purchasing power that can coerce other organizations to follow suit. Only the US military has it. In line with your certification requirements.
This certainly has the potential to mutually accelerate the use of machine learning in materials qualification. For the broader AM sector, therefore, the general process of metal certification can be expected to occur much more quickly than if it were simply allowed to unfold organically without direct government intervention. . The amount of metal required to support accelerated AM metal certification is the largest and most expensive machine, i.e. most parts, should be a major factor in lowering part costs for all forms of hardware in the long run. is. Of course, for this to happen, all the various AI-driven software platforms under development must ultimately be effective in making the process faster and cheaper.
