Page 23 - eBook_Proceedings of the International Conference on Digital Manufacturing V2

Collaborative Study on Weight Optimisation of Lubricant Bottles under
Stacking Condition using Finite Element Analysis and Machine Learning

an overview, Figure 2 provides a schematic representation of
research methodology, presenting the integration of FEA and
Machine Learning in the pursuit of understanding and predicting
the mechanical behaviours of HDPE lubricant cans under stacking
load condition.

Standard Operating Procedure
Finite Element Weight
Analysis Optimization
CAD Model Material Hyperelastic Pre- FEA Parison Structural
Selection Testing Modeling processing Results Controller features
Minimum ASTM Model Selection Boundary
Mass Can D638 through curve fitting Condition Benchmark data Thickness variation
till benchmark
Maximum ASTM Discretization result is achieved
Mass Can D695
STEP 01 STEP 02 STEP 03 STEP 04

Decision Boost SVR
Tree Gradient

FEA simulations Estimated Max Stress
Machine ML Predicted
for data
Learning preparation Training dataset Train ML Model Max. Stress & Results Max
Deformation
Deformation
Parametric Accuracy New Input Data
Analysis 80% Dataset Random Regressor Validation with FEA Results
Bagging
Forest
Tree
Data Hypothesis
splitting
20% Dataset
Note:
Output Features – Maximum Deformation & Stress
Input Features – Variation of bottle thickness across 100 points through Parison Controller Programming (t1-t100)

Figure 2: A schematic diagram of Research Methodology. 4-steps
SOP for weight optimisation of HPDE bottles

The figure shows integration of ML by conducting parametric
study through FEA simulations, considering Maximum Stress and
Maximum Deformation as output features, while bottle wall
7