This paper presents an approach to estimate system parameters of electrical submersible pumps (ESPs) used in the oil and gas industry, employing Physics-Informed Neural Networks (PINNs). ESPs often handle multiphase flows, making accurate modeling crucial for optimizing oil production. However, real-time measurement of fluid and system characteristics is often impractical due to time and cost constraints. The proposed PINN model aims to address this issue by providing an estimation of crucial system parameters using commonly available pressure measurements. The study uses both simulated and experimental data for different water-oil ratios to evaluate the efficacy of the proposed model. The model could potentially reduce the need for expensive field laboratory tests used to estimate fluid properties.
Publication date: 5 Oct 2023
Project Page: https://arxiv.org/abs/2310.03001v1
Paper: https://arxiv.org/pdf/2310.03001
