Elastic modulus distribution measurement of PET bottles based on ultrasonic Lamb waves

However, they lack the capability to provide information about the final room temperature mechanical properties of the produced bottles, which is an important input for subsequent simulations of quality tests such as top load tests, burst tests, and squeeze tests. This limitation has impeded the seamless implementation of computer-aided design in PET bottle production, which is invaluable for reducing trial-and-error in identifying optimal process conditions for specific material or bottle geometry designs.<br><br>These challenges underscore the critical need for an efficient method to measure the mechanical properties of PET bottles.

However, existing mechanical property measurement methods for PET bottles typically involve time-consuming sample preparation steps, constraining their applications. Therefore, the primary objective of this research is to develop an efficient mechanical property (elastic modulus) measurement method for PET bottles and, based on this, to construct a machine learning model capable of predicting the elastic modulus distributions of PET bottles using readily available strain history from process simulations.<br><br>To achieve the objectives, a novel elastic modulus distribution measurement method for PET bottles was proposed in this work, based on the application of ultrasonic Lamb waves.

The effectiveness of this method was validated through tests conducted on stretched PET sheets and PET bottles. Additionally, a machine learning model was developed for predicting the elastic modulus distributions of PET bottles, based on the proposed elastic modulus measurement method. The performance of this machine learning model was then verified through top load simulations of PET bottles.

The work in this thesis provides a reference for the rapid measurement of the mechanical properties of PET bottles and suggests a methodology for real-time quality monitoring during the PET bottle production process. Furthermore, the developed machine learning model lays the groundwork for the virtual optimisation design of PET bottles.<br><br><i>Thesis is embargoed until 31st December 2029</i>.<br><br>