Open Access

ARTICLE

PREDICTION MODEL OF LIQUID HOLDUP BASED ON SOA-BPNN ALGORITHM

by

a The Second Gas Production Plant, PetroChina Changqing Oilfield Company
b Safety and Environmental Supervision Department, PetroChina Changqing Oilfield Company
c Shaanxi Key Laboratory of Advanced Stimulation Technology for Oil & Gas Reservoirs, College of Petroleum Engineering, Xi’an Shiyou
* Corresponding Author. E-mail: 1900679914@qq.com.

Frontiers in Heat and Mass Transfer 2023, 20, 1-6. https://doi.org/10.5098/hmt.20.13

Abstract

In the actual operation of wet gas pipeline, liquid accumulation is easy to form in the low-lying and uphill sections of the pipeline, which leads to a series of problems such as reduced pipeline transportation efficiency, increased pipeline pressure drop, hydrate formation, slug flow and intensified corrosion in the pipeline. Accurate calculation of liquid holdup is of great significance to the research of flow pattern identification, pipeline corrosion evaluation and prediction, and gas pipeline transportation efficiency calculation. Based on the experimental data of liquid holdup in horizontal pipeline, a commonly used BP neural network (BPNN) model is established in this paper. In order to improve the accuracy of BPNN model, Genetic Algorithm (GA), Particle Swarm Optimization (PSO) and Seeker Optimization Algorithm (SOA) are used to optimize the initial weights and thresholds of BPNN model, and GA-BPNN model, PSO-BPNN model and SOA-BPNN model are established. By comparing the model accuracy, the average absolute error of SOA-BPNN prediction model is 3.7351%, and the root mean square error is 0.0113. This model has high prediction accuracy and wide application range, which is obviously superior to other algorithms, and provides a new method for accurate prediction of liquid holdup of wet gas pipeline.

---

Keywords

---