AI-powered Foaming simulation for Prediction and Efficiency

MilanRaval

Introduction

Foaming simulation is a niche technology used by many industries. The utilization of Foaming simulation is increasing similar to casting and other manufacturing simulation. While challenges like trial and error, cost, and time consumption in the manufacturing cycle are solved, additional challenges arose. Challenges like performing multiple foaming simulations by design engineers for decision-making, and simulations that can take days to run on large models (ex. refrigerator foam body, boat, and automotive inside foam fillings.). 

Problem Statement

Physics AI is a fast simulation tool based on physics and frequent design changes.

PolyFoam + Physics AI Workflow

Step 1: Gathering historical data of Foaming simulation performed in the past. Access the h3d files of Injection, Foaming, or Curing stages for training purposes. The below image shows the variable used to create various design iterations for the foaming manufacturing process simulation of refrigerator door cabinets. These design variables consist of different locations of inserts inside the Foam part to analyze density distribution and residual stress analysis.

Step 2: Create a data set for train and testing inside Physics AI. Train Machine Learning algorithm of Physics AI using historical data from Inspire PolyFoam. Define training variables like depth, width, epochs, and a result type that needs to be trained. Here, 16 data sets are taken for training the ML model and 4 are kept aside for testing them against the trained model.

 

Step 3: Test untrained data against trained data to see the predictive accuracy of the Machine learning algorithm. Validate the ML model against the solver through MAE (Mean absolute error). That explains the predictive accuracy of Physics AI. The below image shows the predictability of ML model based on the data used to train it. Physics AI prediction accuracy is pretty accurate with the solver results from traditional foaming simulation.

Step 4: Step 4: Import a new design after the satisfactory MAE is achieved. New design import should be similar in terms of mesh and units for accurate predictions.  Meshed FEM solver input files can be used to import and predict. Before predicting the new design, make sure to activate the machine learning model that is already trained. Then predict a new design. 

Summary: Physics AI is a great tool for fast predictions based on design changes. However, the result's accuracy depends upon the user's trained data set. If the data is not correctly trained, new design predictions won't show promises on result accuracy. So, it is better to understand the type of data being used for training purposes. Below is an example of a new design prediction by Physics AI against actual Foaming simulation results.