Hybrid Digital Twins: Combining Data Analytics and Physics Based Simulation

In the past, customers used to create innovative products by building a hardware prototype of the product, testing the prototype under different operating conditions, refining the prototype, and finally manufacturing the product at scale. That process was long, costly, and error-prone. Increasingly, our customers are using computer-aided design and computer-aided engineering simulation to build software prototypes of the product and performing virtual validation. In the past, the engineering models were not accurate enough to be effective. However, these days with the power of high-performance computing with millions of CPUs and GPUs available on the cloud, we are solving some really large complex models accurately and fast. Furthermore, we are using artificial intelligence, machine learning, and deep learning to accelerate our simulation solvers by factors of 100X. In addition, AI/ML is allowing our customers to use generative design and topology optimization to very rapidly and automatically explore thousands of designs to choose the optimal one. Finally, with the world of connected products and IOT, it is possible to get feedback about how a product is operating in the field, and use that data to build digital twins of the product, and improve the next version of the product. We are working on hybrid digital twins to combine physics-based simulation and data analytics/AI. The future of product innovation in the connected world is very exciting in industries as diverse as high-tech and semiconductors, aerospace and defense, automotive, industrial, and energy.