Are complex black-box models for Permanent Downhole Gauge pressure estimation necessary?

作     者：Huffner, Leonardo Nardi Trierweiler, Jorge Otavio Farenzena, Marcelo 

作者机构：Fed Univ Rio Grande Sul UFRGS Grp Intensificat Modeling Simulat Control & Optim Chem Engn Dept R Engn Luiz Englert S-NCampus Cent BR-90040040 Porto Alegre RS Brazil 

出 版 物：《JOURNAL OF PETROLEUM SCIENCE AND ENGINEERING》 (石油科学和石油工程杂志)

年 卷 期：2019年第173卷

页      面：715-732页

核心收录：

学科分类：0820[工学-石油与天然气工程] 08[工学] 

基　　金：National Petroleum, Natural gas and Biofuels Agency (ANP) Petrobras Brasileiro S.A. - PETROBRAS 

主　　题：Subsea plant models model identification Batch data processing Production systems BLACK-BOX TRANSFER FUNCTIONS Temperature Sensor Device Component PHGDH gene 

摘      要：The Permanent Downhole Gauge (PDG) is a pressure and temperature sensor located subsea near the perforation point in the offshore oil production system. This sensor is very useful in operation problems detection (fouling, valves, plugging, etc.), multiphasic flow analysis, production tests adjustment, control strategy and model identification. Given its location under hazardous conditions, it may come to failure or imprecisions in measurement. The high costs related to the maintenance make it infeasible to perform. To overcome this issue, a methodology based on digital signal processing (DSP) is proposed in order to design a low pass digital filter aiming the PDG pressure reconstruction. The proposed methodology requires no plant model, low computational cost and only the Christmas-Tree pressure measurement. The proposed methodology is compared with two commonly employed black-box methodologies: Neural Network and Transfer Function Estimation. The results show the success of the designed filter and the advantages over the black-box approaches. The estimation of the PDG pressure from three operating wells using real process data was performed.