Training Surrogate Models to Build Faster Simulation Apps and Digital Twins

Duration: 27:03

The use of simulation-data-driven surrogate models in simulation apps yields compact apps that give near-instantaneous results for a seamless user experience. In this archived webinar, we discuss how to build faster simulation apps and digital twins with surrogate models using the COMSOL Multiphysics® software.

To start, we share how one company is using simulation apps to predict curing times for concrete castings and how this app could be made even more powerful with new surrogate model functionality released in version 6.2 of COMSOL Multiphysics®. Next, we discuss the surrogate model workflow, including generating the training data, using the surrogate model functions, defining the deep neural network (DNN) layers, and more.

The webinar concludes with a step-by-step demonstration on how to build an app of a tubular reactor that uses a surrogate model to provide quicker evaluations.

Next Step

Download the examples via Resources on Training Surrogate Models for Faster Simulation Apps.