Simulation has truly revolutionized the way we approach solving complex problems and building powerful products. Engineers & designers can simulate various scenarios and test their hypotheses to achieve desired outcomes without the need for costly physical prototypes & experimental spaces. With the increasing complexity of today’s problems, the incorporation of future technologies is necessary to evolve with changing demands & times. One such technology is Artificial Intelligence (AI) which helps us better understand the outcomes of simulations by analyzing vast amounts of data and predicting scenarios with valuable insights that may not be readily discernible through traditional methods.
Here are few highlights we found from talking to many simulation experts over the past year:
- Simulation users spend an excessive amount of time fixing geometries. I know this is not believable, but we found this across the board.
- Analysts and design engineers who use simulation take a reasonable time to set up the simulation (also called the pre-processing stage). We asked users if they would like to cut down this time. They replied with a resounding YES.
- Meshing failures, bad elements, and lack of good quality mesh in critical areas are more common than you think. This finding surprised us and is connected strongly to the first point about bad geometry.
- Simulation users spend less time analyzing results than they would like to. It is because they are well behind on their deadline when they get to analyzing results. This was the most critical finding, in my opinion.
Simulation solution providers can offer help only if they acknowledge these ground realities. So, it is only natural to expect AI first to help simulation users in these four areas.
The Incorporation of Artificial Intelligence in Simulation
Altair has started on the right path to empower simulation users by employing AI to handle the non-productive part of their job. Altair talked about this during their “Design Explorer: Driving the Simulation Process using Artificial Intelligence” webinar from the at Altair Future Industry 23 event.
Associate Analyst Sarthak Sheladia shared their thoughts about Altair and the incorporation of AI in simulation.
Foremost, let us understand how the platform works. Design Explorer is a beginning-to-end design ideation workflow that provides multi-run simulation setup, execution, and interpretation. It empowers teams to break the typical iterative cycle of conducting simulations, checking results, & making adjustments by giving a complete evaluation of design possibilities for their products.
Including AI in Design Explorer is a natural fit because various datasets generated during design exploration can be trained with machine learning algorithms to expose powerful AI prediction capabilities. Teams can explore the true breadth of design possibilities for their products with the support of physics AI tools such as scalar prediction, field prediction, & curve prediction and analyze datasets & model changes in real-time. This eliminates the need to test different solvers for desirable designs & human error, bringing new innovative ideas to life.
During these initial stages, the role of AI in the design workflow process may feel in the experimental settings as the models are still being trained on design explorer exploration data. Additionally, engineers still want a lot of manual ownership over the simulation inputs governed by the model. But as models become more accurate in predicting simulation changes, teams have an advantage of quick decision-making and newly inspired designs for their products which can tackle bolder challenges.
What’s my take?
Along with other industries, Engineering is at the doors of AI revolution. What seemed like just a concept of the future, may well be our reality soon. Engineers will have access to plethora of tools to build products that they want and creating a niche of early adopters might be the key for many companies. Altair has made the right choice with incorporation of AI into its underlying technologies to deliver on big promises for the future of simulation.
Should AI be used in transforming simulation? No question about that, but not all areas deliver the same bang for the buck.