Degradation Mechanism Study of FEPM Elastomer in HPHT Sour Environments and Material’s Service Lifetime Prediction by Using a Physics-Informed Neural-Network Modeling Method

Abstract Emerging deep-well drilling and completion operations require sealing materials to withstand high temperature high pressure (HPHT) and sour downhole environments. Tetrafluoroethylene-propylene elastomer (FEPM) was widely used in the industry for such harsh environments. In the work, the degradation mechanism of FEPM in the condition was thoroughly studied. The chemical structure change was correlated with the physical properties change. Furthermore, a physics-informed neural network (PINN) model was developed to predict the material’s service life in the condition. Two grades of FEPM rubbers were selected to be aged in the ISO 23936-2 high H2S condition at 200, 225, 250◦C for various durations as 7, 14, 28, and 42 days. After each duration, the samples were taken out from autoclaves for tensile, Dynamic Mechanical Analysis (DMA) testing. In addition, Fourier-transform Infrared Spectroscopy (FT-IR) and solid state Nuclear Magnetic Resonance (NMR) measurements were performed to analyze the chemical structure change of the materials during the aging process. A PINN model was developed and trained with experimental data to predict the material’s service life under different conditions. The tensile strength decreased, while the elongation at break increased after aging 7 days in comparation with the properties of the sample in wet conditions. The tensile strength showed a general trend of reduction by increasing temperature and time. The FT-IR and solid-state NMR results displayed that the carbonyl group amount was reduced at the early stage of aging. However, the reduction trend slowed down as aging time increased. The results implied the early reduction of the physical properties was partially due to the crosslinking degree reduction. FT-IR and solid-state NMR results show that the other functional groups did not change much at the longer aging time. The results implied that the long-term physical properties change could be due to the absorption of hydrocarbons and H2S by elastomers as well as the slow reduction of crosslinking density. A PINN model was developed to capture the constitutive properties: elongation at break, tensile strength, and 50% modules of the materials at various aging durations. It is shown that the PINN model can predict the FEPM material’s service life better than the Arrhenius equation. To the best of our knowledge, the proposed model is the first machined learned tool to predict the service life of the elastomers, which makes the work the first comprehensive study on using the degradation mechanism of FEPM elastomers in HPHT sour condition to train a hybrid physics-driven acritical intelligence modeling method. The method displayed a significant advantage compared with the industry-widely used Arrhenius equation. These results could help the industry to better understand the sealing material’s behavior in the HPHT sour environment in the long term.