Building Digital Twins at Scale

Sandeep Urankar - Rescale

Abstract:

Abstract Industry 4.0 is rapidly transforming manufacturing through a convergence of technologies like High-Performance Computing (HPC) and Artificial Intelligence/Machine Learning (AI/ML). These converging technologies unlock new use cases for Digital Twins in digital engineering and manufacturing. Digital twins unlock predictive analytics, allowing manufacturers to accelerate optimal designs, identify defects, improve process efficiency, and reduce capital and operating costs. Digital Twins can be used to generate insights into operational data or simply to speed up the design process. In either case, the foundational need is to develop a machine-learning model or surrogate model. This involves extensive design space exploration, where various parameters within physics-based models are systematically varied. Understanding a system's range of behaviors and limitations is critical for training robust and reliable AI/ML models. The process is computationally intensive and generates large amounts of data, necessitating effective management.

In this presentation, we cover:

1. How organizations flexibly combine best-in-class digital engineering tools in the cloud Digital Twins.

2. How organizations leverage the latest computational architectures to accelerate R&D breakthroughs.

3. How design exploration can be automated to build effective ML models that generate improved product designs faster