COMSOL, a leading simulation software environment, will host a significant conference in Burlington, Massachusetts, aimed at advancing the field of multiphysics simulations. This week’s TNC will be no exception, it will be filled with engaging activities! Combined with inspiring keynotes, hands-on workshops and product demonstrations, attendees will leave empowered to conquer the challenge of simulation time constraints.
The conference will unite civil and military engineers, academic researchers, and industry practitioners. Join them to discover how COMSOL’s new surrogate models are transforming engineering workflows, particularly in the automotive industry. These approaches enable more efficient and faster modeling of EV battery modules. Equipped with this power, producers can take quicker and more impactful data-driven decisions that increase their operational effectiveness.
COMSOL’s parent company, headquartered in Stockholm, have developed these surrogate models. In doing so, they have realized amazing computational speedups on the order of 100,000x faster than previous methods. Through a semi-supervised approach, this technology uses neural networks alongside traditional mathematical reduced order modeling (ROM) tools enabling real-time assessments.
“Surrogate models are a really promising new technology,” remarked Bjorn Sjodin, senior vice president of product management at COMSOL. “They let you take a detailed, full multiphysics model and distill that down into something much simpler and faster to evaluate that you can use machine learning to build.”
In fact, European automotive manufacturers have already started implementing these surrogate models into their design workflows. By making simulation processes more agile, they’re able to move much more quickly with the changing demands of EV production.
Recently, a Swiss academic institute has deployed COMSOL’s surrogate system as a customized app tailored for use by Indian farmers. Despite the hype, this development extends far past automotive applications. This app, a first of its kind, uses predictive analytics to forewarn food spoilage in warehouses, eventually giving farmers more control over perishable produce. One group reported a 20 percent decrease in spoilage through the use of this app. This outcome serves to illustrate the versatility and overall transformative impact of COMSOL’s technologies across several sectors.
When you get to factory floor and you’d like to pass this knowledge on, these surrogate models are wildly valuable. They allow for rapid assessments and on-the-spot outcomes,” Sjodin continued.
The surrogate modeling system specifically allows users to package their apps into standalone executables. These executables can then be freely distributed around the world with no licensing costs. This kind of flexibility unlocks exciting opportunities for organizations looking to harness the power and promise of advanced simulation technologies, without the steep price tags.
COMSOL’s complete model with all unknown inputs normally takes more than 15 minutes to solve. Engineers have the opportunity to adopt surrogate models as a practice enabler more broadly. This often enables them to skip most of the downtime, massively increasing productivity and efficiency within engineering workflows.
Throughout the conference, participants are able to immerse themselves in hands-on workshops and demonstrations. These workshops will highlight real-world uses of these cutting-edge modeling techniques. On the third day, participants will debate the vital role surrogate models play in carrying innovations beyond the lab and into urgent challenges within engineering simulations.
“ROMs are divided into two big families: intrusive methods, in which one manipulates directly the governing equations, and non-intrusive methods, in which only the simulation data are considered,” – researchers.