Near-instant physics predictions saves millions of dollars

About the customer

Since 1988, Kinetic Vision, a United States (U.S.)-based design and development consulting firm, has partnered with Fortune 500 companies to create transformative products, processes and digital-first solutions. These partnerships span multiple vertical markets, including consumer packaged goods (CPG), medical technology, automotive, aerospace, supply chain and more. The Kinetic Vision team includes industry experts on using the Digital Twin for developing and optimizing products, systems, processes and operations. These multidisciplinary experts combine the latest digital tools, like artificial intelligence (AI), with extensive prototyping, testing and metrology capabilities.

Kinetic Vision’s CPG customers face constant pressure to reduce packaging costs and improve sustainability metrics without compromising on performance. According to the United Nations University Institute for Water, Environment and Health, the world produced 600 billion plastic bottles in 2021. Environmentally speaking, this makes bottles an obvious target for material minimization. Savings of 20 to 30 percent are typical for lightweighting efforts. This would be the equivalent of taking billions of plastic bottles out of landfills annually. However, traditional finite element analysis (FEA) studies are time and resource intensive, often requiring weeks or months to arrive at optimum designs. This limits the ability of packaging engineers to explore alternative geometries at-scale, leaving billions of dollars (according to Future Market Insights’ Blow Molded Plastic Market 2024 report) in potential material savings untapped.

Top-load simulations showing stress distribution (left) and crush test rendering (right) for a lightweighted bottle design.

Our solution

Kinetic Vision adopted Simcenter™ PhysicsAI™ software, a breakthrough AI-driven simulation approach that combines the power of deep learning and historical computer-aided engineering (CAE) data to deliver fast physics predictions. By training Simcenter PhysicsAI with historical simulation data, Kinetic Vision was able to replace lengthy FEA runs with near-instant predictions, cutting solution times from hours or days to seconds.

To validate solution accuracy, Kinetic Vision compared Simcenter PhysicsAI results to those from conventional simulation methods for four nonlinear scenarios.

Data concerning the simulation model scenarios and the training of their corresponding Simcenter PhysicsAI models are summarized below:

1. Modeling the fit of a fabric on a walking human.
   • Simulation: A dynamic, explicit model run using LS-DYNA
   • Computations: Solved in 24 hours using 64 high-performance computer processing units (CPUs)
   • Simcenter PhysicsAI model trained on 22 simulation results
   • Training time: Took 2.5 hours using a traditional CPU
2. Assessing the second and third phase buckling of a component.
   • Simulation: A dynamic, explicit model run using Abaqus
   • Computations: Solved in 2 hours using 64 high-performance CPUs
   • Simcenter PhysicsAI model trained on 180 simulation results
   • Training time: Took 48 hours using a traditional CPU
3. Optimizing the blow molding process to make a bottle.
   • Simulation: A dynamic, explicit model run using Abaqus
   • Computations: Solved in 7 hours using 64 high-performance CPUs
   • Simcenter PhysicsAI model trained on 47 simulation results
   • Training time: Took 9 hours using a traditional CPU
4. Evaluating the stresses and deformations from crushing a bottle with a top load.
   • Simulation: A dynamic, explicit model run using Abaqus
   • Computations: Solved in 35 minutes using 64 high-performance CPUs
   • Simcenter PhysicsAI model trained on 45 simulation results
   • Training time: Took 7 hours using a traditional CPU

The resulting computational times and accuracy of the Simcenter PhysicsAI models, compared to traditional simulations, is displayed in figure 1. The graph shows order of magnitude improvements for computational speeds. AI results also show high accuracy when compared to FEA results. With this knowledge, product lightweighting for Kinetic Vision’s main CPG customers was reduced from weeks to days, enabling the customer brands to save millions of dollars annually, while dramatically reducing plastic waste.

A corresponding Simcenter PhysicsAI model produces results 350 times faster with only a 2% error.

Results

Kinetic Vision’s product packaging lightweighting study revealed how:

• CPG organizations can save millions of dollars annually by minimizing material usage across multiple high-volume production lines
• The equivalent of billions of bottles can be eliminated from landfills each year using lightweight, sustainable design
• Simcenter PhysicsAI models using standard CPUs can run 350 to 4,000 times faster than traditional FEA while maintaining accuracy in the range of 87.5 to 97.5 percent
• AI produced results can match the engineering rigor of simulations while scaling innovation

“Lightweighting isn’t just about cutting costs, it’s about multiplying impact,” says Shane Mooney, director of development at Kinetic Vision. “Every gram removed saves resin, reduces waste and scales to millions of units on store shelves. Using Simcenter PhysicsAI makes it possible to deliver these results faster than ever before.”

Plastic blow molded products are everywhere, and most are conservatively designed from legacy prototypes and test methodologies, even though validated simulation methods have existed for decades. The hundreds of simulations and associated weeks, or months, needed to optimize CPG designs are seen as unnecessary impediments to rapid market introduction and product sales.

Leveraging Simcenter PhysicsAI changes that. The result is that leading companies can save hundreds of millions of dollars annually while removing the equivalent of billions of plastic bottles from landfills. For example, by unlocking rapid design iteration, using Simcenter PhysicsAI empowered Kinetic Vision and its CPG partners to deliver lighter, stronger and more sustainable product packaging at unprecedented speed.

Kinetic Vision Innovation Center in Cincinnati, Ohio.