Machine Learning Potential-Driven Investigation of NEPE Matrix: Mechanical Properties and Failure Mechanism.

Abstract

The mechanical properties of nitrate ester plasticized polyether (NEPE) propellant are critical in determining the safety and performance of solid rocket engines. However, understanding the failure mechanisms at atomic to micron scales remains a persistent challenge. In this work, we present the first application of machine learning potential (MLP) for NEPE, achieving ab initio-level accuracy while significantly enhancing computational efficiency and accuracy compared to traditional methods. Using the MLP model, molecular dynamics (MD) simulations were conducted to investigate the effects of molecular size, strain rate, and temperature on the mechanical behavior of NEPE. Key findings indicate that the mechanical performance is highly sensitive to temperature fluctuations, with tensile strength decreasing significantly from 240 to 330 K. To bridge the gap between MD simulations and experimental results, the time-temperature superposition (TTS) principle was employed, enabling a reliable virtual evaluation of the mechanical properties of NEPE matrix. The predicted tensile strength range of 8 to 22 MPa aligns well with experimental data, validating the proposed approach. This research not only enhances the understanding of the mechanical properties of NEPE at the atomic level but also establishes a robust framework for high-performance propellant design through the integration of machine learning potentials and multiscale modeling techniques. The findings provide valuable insights for optimizing the safety and functionality of NEPE in solid rocket applications.