Page 37 - 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

approximately 50 minutes to obtain the results from single FEA
simulation.

Table 4: Comparison of FEA Results with ML Outputs

Output Features
Input Feature % Error
ML FEA
t1- t11- t21- t31- t41- t51- t61- t71- t81- t91- Max. Max. Max. Max. Max.
10 20 30 40 50 60 70 80 90 100 Def. Max. Stress Def. Stress Def Stress
Mm mm mm mm mm mm mm mm mm mm mm MPa mm MPa
1.0 1.0 1.0 1.0 1.0 1.0 1.0
1.17 1.0 1.17 2.026 17.577 2.09 18.50 3.06 4.99
1.17 1.17 1.17 1.17 1.0 1.0 1.0 1.0 1.0
1.17 1.716 15.410 4.06 5.65
1.786 16.333

CONCLUSION AND FUTURE WORK

The collaborative study aimed at developing a Standard Operating
Procedure for the weight optimisation of HDPE bottles under
stacking conditions has been conducted using Finite Element
Analysis and Machine Learning. Two possible approaches have
been introduced to optimise the overall mass of the maximum
bottle using FEA methodologies. The FEA results reveal a
substantial reduction in the bottle’s mass through the
implementation of structural reinforcement features and parison
controller programming, leading to significant material cost
savings while maintaining the bottle’s structural integrity.
Furthermore, this study presents a novel approach utilising
Machine Learning as a supplementary tool for Finite Element
Modelling to predict the Hyper elastic behaviour of High-Density
Polyethylene, aiming to replace highly iterative, computationally
expensive and time-consuming simulations by employing a
standard machine learning algorithm.

To achieve this, five different ML models have been trained
using data prepared from parameterised simulations and the
thickness of the bottle profile is systematically varied along its
length via parison controller programming. The performance of
these models has been evaluated using two metrics: Mean Square
Error and R² score. The results indicate that the Random Forest

21