Using Digital Twin for faster & accurate performance prediction of Industrial Assets.
2022-26-1146
05/26/2022
- Event
- Content
- Digital Twin is an emerging trend on Industrial use cases. Digital Twin is a Replica of a physical asset in Real time. The traditional way of developing Digital twin is, by using complex transfer functions. However, this requires tremendous amount of expertise with good understanding of the domain & the physics. The accuracy of the Twin model is directly dependent on the transfer functions, which many times result in a disconnect with the physical world. Physics based simulations can result in a better accuracy. However, the time taken by the software to make Real time virtual simulations is a bigger Industrial challenge due to high computational speed. So, we need to strike the right balance between speed & accuracy. There are some software who are attempting to provide solutions to this challenge. One of the techniques is to use Reduced Order Modelling (ROM) to speed up the computation of virtually simulating complex Industrial assets, without too much of approximation. The intent of this paper is to describe the possible ways of building a ROM & how it helped us to solve some industrial challenges such as • To speed up visualizing the performance parameters of physical industrial assets on the field with Real-time sensor data, thereby giving accurate inputs to the predictive analytic engine. • To reduce the decision making time of accepting / rejecting the manufacturing non-conformities, reported from the inspection line.
- Citation
- Narasimhan, V., "Using Digital Twin for faster & accurate performance prediction of Industrial Assets.," SAE Technical Paper 2022-26-1146, 2022, .