Page 18 - eBook_Proceedings of the International Conference on Digital Manufacturing V2

Proceedings of the International Conference on Digital Manufacturing –
Volume 2

A dataset with 100 inputs (bottle thickness across 100 points)
and outputs features (maximum stress and deformation) are used
to train five machine learning algorithms – Support Vector
Regressor, Random Forest Regressor, Gradient Boosting
Regressor, Decision Tree Regressor, and Bagging Regressor.
They are evaluated based on Mean Squared Error and R-squared
metrics, with FEA results used for validation. The redesigned
bottles show a potential weight reduction of 25.71% compared to
the reference design. Among five models, Random Forest
Regressor demonstrates the best performance, achieving lower
MSE values of 0.17993 and 0.00213 for stress and deformations,
coupled with higher R-squared scores of 0.992 and 0.995. This
study shows the potential of Machine Learning to achieve
lightweight structure design, reducing reliance on FEA for stress
analysis applications.

Keywords: Hyper-elastic Modelling, Lightweight design,
Machine Learning

INTRODUCTION

Polymer such as high-density polyethylene (HDPE) has widely
been used for the storage containers of oil or milk due to its good
mechanical properties, being certified food grade, and having low
manufacturing cost. In addition, virgin HDPE can also be recycled
which makes it an appropriate choice to be used by manufacturing
companies for such highly disposable products. However, the
performance of a plastic container depends on how it is made and
stored. Figure 1 illustrates the potential causes linked with the
unconditional damage of containers which could happen due to
stacking practices in terms of uneven load distribution and
excessive carton height, buckling or bulging of bottom of the
bottles due to prolong stacking (de Queiroz, Silveira, de Camargo
& Vaccari, 2018), effect of creep under sustained stress, failure of
ASTM standard related to pressure testing and drop testing, and
dynamic load due to shock and vibration during transportation that
have commonly been reported, resulting in significant economic
losses to the manufacturer.

2