Machine leaning aided accelerated characterization of temperature and strain rate-dependent dynamic properties of close-cell polymer foams

Abstract

Dynamic mechanical analysis (DMA) is commonly used to test the viscoelastic properties of materials. However, the frequency-dependent component of the measured properties in such techniques remains underutilized due to the lack of correlation between frequency, temperature, and strain rate. Recent studies have used a time-temperature superposition (TTS) based method to develop a frequency to time domain transformation and validated it for tensile loading conditions. This transform is very useful in reducing the experimental effort required to measure material modulus over a wide range of temperatures and strain rates. However, this transform is not yet validated on an important class of materials, closed-cell foams. The present work is focused on understanding the validity of such transforms on A polymethacrylimide based close-cell structural foam, Rohacell®. Torsional properties of these foams are important for underwater structural applications but neither these foams have been studied for such properties nor the transforms have been tested for their validity for such loading conditions. DMA was conducted under tension and torsion over 30-210°C and 0.1–10 Hz. The transformed results validated against tensile test results revealed an error <6.63% in the strain rate range of ${10}^{-5}$ to ${10}^{-2}$ s^-1. Moreover, the shear modulus is also transformed from frequency to time domain using the same scheme. Combining machine learning, the results yield an error <4.71% in the shear strain rate of ${10}^{-4}$ and ${10}^{-3}$ s^-1. These results show the validity of the transformation method as well as present valuable information about the mechanical properties of Rohacell.