HP and NVIDIA Collaborate on Open-Source Manufacturing Digital Twin

HP 3D Printing and NVIDIA Modulus have introduced a collaboration to develop an open-source manufacturing digital twin, leveraging physics-informed machine studying (physics-ML). This partnership goals to foster innovation in AI engineering purposes by embedding bodily legal guidelines into the educational course of, in line with NVIDIA Technical Weblog.

Developments in Physics-ML

Physics-ML is a burgeoning area that includes bodily legal guidelines into machine studying fashions, enhancing the generalizability and effectivity of neural networks. NVIDIA Modulus, an open-source framework, facilitates the constructing, coaching, and fine-tuning of those fashions with a easy Python interface. The framework gives reference purposes to assist area specialists apply physics-ML to real-world use circumstances.

The Digital Twin workforce at HP 3D Printing Software program Group has utilized physics-ML fashions for his or her manufacturing digital twin and contributed this work to Modulus. HP, a frontrunner in additive manufacturing, goals to speed up the onboarding of recent purposes and undertake this expertise in manufacturing environments. Dr. Jun Zeng, HP’s distinguished technologist, emphasised the significance of physics simulation engines grounded in manufacturing course of variability, noting the numerous speedups achieved with well-trained physics-ML fashions.

Digital Twins in Additive Manufacturing

HP has a wealthy historical past of technological innovation, together with the event of thermal inkjet expertise. The corporate’s newest innovation, HP Steel Jet, permits the manufacturing of industrial-grade 3D steel components. HP is creating a digital twin for Steel Jet expertise to optimize design parameters and course of management, thereby enhancing half high quality and manufacturing yield.

The HP workforce created the Digital Foundry Graphnet mannequin, making use of physics-ML to speed up the computation of steel powder materials part transitions. This mannequin has achieved important speedups, enabling close to real-time, high-fidelity emulation of the steel sintering course of. The mannequin has additionally demonstrated its applicability to various geometrical designs and course of parameter configurations.

Physics-ML Innovation at HP

Though physics-ML remains to be in its early phases, the HP Digital Twin workforce believes within the position of the open-source group in accelerating its improvement. By open-sourcing Digital Foundry Graphnet by means of NVIDIA Modulus, HP has joined the physics-ML open-source group. Conventional high-fidelity physics simulations are computationally intensive, usually taking hours or days for one design iteration. Physics-ML surrogate fashions provide high-fidelity emulation, enabling quicker design iterations.

Immediate suggestions on product design manufacturability and automatic design screening at the moment are potential with physics-ML surrogate fashions. These fashions additionally enable product design groups to make use of prior simulation knowledge as a supply of ground-truth knowledge. The combination of product design and manufacturing optimizations, which historically required a number of iterations between departments, can now be considerably accelerated.

HP’s course of physics simulation software program, Digital Sintering, has been deployed to HP Steel Jet clients to enhance manufacturing outcomes. Operating a well-trained steel sintering inferencing engine takes simply seconds to acquire the ultimate sintering deformation worth, considerably decreasing the time required for design iterations.

Empowering Researchers

Physics-ML surrogate fashions are on the forefront of near-real-time simulation workflows. Improvements like Digital Foundry Graphnet show the facility of AI to speed up simulation workflows, delivering predictions in seconds. Democratizing AI for manufacturing is crucial to empower a wider vary of innovators to resolve trade challenges.

AI researchers and the HP 3D Printing workforce make the most of the NVIDIA Modulus open-source challenge to collaborate with area specialists. NVIDIA helps the physics-ML analysis group by offering a platform that enhances collaboration and innovation, guaranteeing that superior AI instruments are accessible to all.

Picture supply: Shutterstock